Distress under rescue rule S3, k=2. generalisation check, gap 0.453. 90.5 % of clips score at or below zero on this emotion and the largest gap on its normalised axis is 0.453 (WIDER than the 0.25 step cap). This rule found 7,875 chains over 40,000 tracks; the strict rule found 0 at k=3.
S3 changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'. What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained. Full explanation →Interest has a median of 2.08 and is never zero, while Infatuation is zero on 87.7 % of clips — and the caption named an emotion whenever its raw score cleared an absolute 1.0. Interest therefore appeared in 94.8 % of captions and Sadness in almost none: the clause was reporting the scale of the head, not the emotion of the clip. An emotion is now named only when it lands in the top 10 % for that emotion, against a pooled tie-aware mid-rank ECDF over 132,833,726 utterances spanning every dataset and language. Interest now appears in 6.2 %, all 40 emotions occur, and a clip that is ordinary on all 40 says “no dominant emotion” rather than being forced to pick one (21.8 % of clips). This is the same scale the trajectory miner selects on, so the caption and the mining now refer to the same quantity: the mined target emotion is named in the final clip's caption on 73 % of chains, up from 46 %.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.19. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.19.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.911 before conversion and 0.880 after — it fell by 0.031. Neighbour-to-neighbour the worst pair went 0.911 → 0.880. (The earlier render, with segment 1 left raw, scores 0.865 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.545 in the original and +0.536 after conversion — 98 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.87 → 3.23 (+0.36) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.02, 1.02. In the first clip the scorer found no Distress whatsoever (0.02); by the last it is at 1.02.
On the corpus-wide percentile scale those become 0.89, 0.98 — a total move of +0.09.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.695 before conversion and 0.780 after — it rose by 0.084. Neighbour-to-neighbour the worst pair went 0.695 → 0.780. (The earlier render, with segment 1 left raw, scores 0.689 here.) This chain started 0.50-0.70 — audibly different, a band where the conversion is close to a wash on this measure.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.086 in the original and +0.525 after conversion — 612 % of the delta retained, i.e. the move came out slightly larger after conversion than before.
Quality. Mean predicted overall quality across the segments went 2.75 → 3.07 (+0.33) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.47. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.47.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.304 before conversion and 0.488 after — it rose by 0.185. Neighbour-to-neighbour the worst pair went 0.304 → 0.488. (The earlier render, with segment 1 left raw, scores 0.392 here.) This chain started below 0.50 — the segments really were different people, the band the conversion helps most: chains starting below 0.50 improve on this measure almost without exception.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.552 in the original and +0.533 after conversion — 97 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.68 → 3.06 (+0.37) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.02. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.02.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.760 before conversion and 0.841 after — it rose by 0.081. Neighbour-to-neighbour the worst pair went 0.760 → 0.841. (The earlier render, with segment 1 left raw, scores 0.841 here.) This chain started 0.70-0.80 — close, but under the identity threshold, a band where the conversion is close to a wash on this measure.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.537 in the original and +0.540 after conversion — 101 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.75 → 3.21 (+0.45) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.24. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.24.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.741 before conversion and 0.616 after — it fell by 0.125. Neighbour-to-neighbour the worst pair went 0.741 → 0.616. (The earlier render, with segment 1 left raw, scores 0.369 here.) This chain started 0.70-0.80 — close, but under the identity threshold, a band where the conversion is close to a wash on this measure.
The emotional move did not survive. Re-scored end to end, Distress moved +0.546 in the original and -0.043 after conversion — it changed direction. On this chain the corrected audio is not an improvement.
Quality. Mean predicted overall quality across the segments went 2.45 → 2.85 (+0.40) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.12. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.12.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.802 before conversion and 0.798 after — it fell by 0.004. Neighbour-to-neighbour the worst pair went 0.802 → 0.798. (The earlier render, with segment 1 left raw, scores 0.511 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.542 in the original and +0.061 after conversion — 11 % of the delta retained, so a meaningful part of the trajectory was flattened.
Quality. Mean predicted overall quality across the segments went 2.46 → 2.82 (+0.37) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.36. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.36.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.954 before conversion and 0.839 after — it fell by 0.115. Neighbour-to-neighbour the worst pair went 0.954 → 0.839. (The earlier render, with segment 1 left raw, scores 0.872 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.550 in the original and +0.548 after conversion — 100 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.73 → 3.03 (+0.30) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.00. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.00.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.303 before conversion and 0.802 after — it rose by 0.499. Neighbour-to-neighbour the worst pair went 0.303 → 0.802. (The earlier render, with segment 1 left raw, scores 0.801 here.) This chain started below 0.50 — the segments really were different people, the band the conversion helps most: chains starting below 0.50 improve on this measure almost without exception.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.536 in the original and +0.496 after conversion — 92 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.95 → 3.39 (+0.44) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.08. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.08.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.608 before conversion and 0.714 after — it rose by 0.106. Neighbour-to-neighbour the worst pair went 0.608 → 0.714. (The earlier render, with segment 1 left raw, scores 0.716 here.) This chain started 0.50-0.70 — audibly different, a band where the conversion is close to a wash on this measure.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.540 in the original and +0.000 after conversion — 0 % of the delta retained, so a meaningful part of the trajectory was flattened.
Quality. Mean predicted overall quality across the segments went 2.47 → 2.85 (+0.38) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.11. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.11.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.938 before conversion and 0.827 after — it fell by 0.110. Neighbour-to-neighbour the worst pair went 0.938 → 0.827. (The earlier render, with segment 1 left raw, scores 0.752 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.541 in the original and +0.000 after conversion — 0 % of the delta retained, so a meaningful part of the trajectory was flattened.
Quality. Mean predicted overall quality across the segments went 2.85 → 3.11 (+0.26) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.02. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.02.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.907 before conversion and 0.841 after — it fell by 0.066. Neighbour-to-neighbour the worst pair went 0.907 → 0.841. (The earlier render, with segment 1 left raw, scores 0.811 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.537 in the original and +0.000 after conversion — 0 % of the delta retained, so a meaningful part of the trajectory was flattened.
Quality. Mean predicted overall quality across the segments went 3.03 → 3.15 (+0.12) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.15. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.15.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.793 before conversion and 0.793 after — it fell by 0.000. Neighbour-to-neighbour the worst pair went 0.793 → 0.793. (The earlier render, with segment 1 left raw, scores 0.683 here.) This chain started 0.70-0.80 — close, but under the identity threshold, a band where the conversion is close to a wash on this measure.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.543 in the original and +0.554 after conversion — 102 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.85 → 3.00 (+0.15) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.24. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.24.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.166 before conversion and 0.533 after — it rose by 0.366. Neighbour-to-neighbour the worst pair went 0.166 → 0.533. (The earlier render, with segment 1 left raw, scores 0.453 here.) This chain started below 0.50 — the segments really were different people, the band the conversion helps most: chains starting below 0.50 improve on this measure almost without exception.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.546 in the original and +0.000 after conversion — 0 % of the delta retained, so a meaningful part of the trajectory was flattened.
Quality. Mean predicted overall quality across the segments went 2.70 → 2.84 (+0.14) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.02. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.02.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.081 before conversion and 0.434 after — it rose by 0.353. Neighbour-to-neighbour the worst pair went 0.081 → 0.434. (The earlier render, with segment 1 left raw, scores 0.349 here.) This chain started below 0.50 — the segments really were different people, the band the conversion helps most: chains starting below 0.50 improve on this measure almost without exception.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.537 in the original and +0.528 after conversion — 98 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.76 → 3.01 (+0.25) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.33. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.33.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.875 before conversion and 0.814 after — it fell by 0.061. Neighbour-to-neighbour the worst pair went 0.875 → 0.814. (The earlier render, with segment 1 left raw, scores 0.754 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.549 in the original and +0.534 after conversion — 97 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 3.14 → 3.33 (+0.19) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.06. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.06.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.448 before conversion and 0.827 after — it rose by 0.379. Neighbour-to-neighbour the worst pair went 0.448 → 0.827. (The earlier render, with segment 1 left raw, scores 0.659 here.) This chain started below 0.50 — the segments really were different people, the band the conversion helps most: chains starting below 0.50 improve on this measure almost without exception.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.539 in the original and +0.506 after conversion — 94 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.86 → 3.16 (+0.30) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.85. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.85.
On the corpus-wide percentile scale those become 0.44, 1.00 — a total move of +0.56.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.267 before conversion and 0.774 after — it rose by 0.507. Neighbour-to-neighbour the worst pair went 0.267 → 0.774. (The earlier render, with segment 1 left raw, scores 0.715 here.) This chain started below 0.50 — the segments really were different people, the band the conversion helps most: chains starting below 0.50 improve on this measure almost without exception.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.556 in the original and +0.501 after conversion — 90 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.88 → 3.08 (+0.20) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.36. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.36.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.586 before conversion and 0.755 after — it rose by 0.170. Neighbour-to-neighbour the worst pair went 0.586 → 0.755. (The earlier render, with segment 1 left raw, scores 0.546 here.) This chain started 0.50-0.70 — audibly different, a band where the conversion is close to a wash on this measure.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.550 in the original and +0.000 after conversion — 0 % of the delta retained, so a meaningful part of the trajectory was flattened.
Quality. Mean predicted overall quality across the segments went 2.69 → 2.95 (+0.26) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.50. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.50.
On the corpus-wide percentile scale those become 0.44, 0.99 — a total move of +0.55.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 100–100 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.627 before conversion and 0.689 after — it rose by 0.062. Neighbour-to-neighbour the worst pair went 0.627 → 0.689. (The earlier render, with segment 1 left raw, scores 0.658 here.) This chain started 0.50-0.70 — audibly different, a band where the conversion is close to a wash on this measure.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.553 in the original and +0.552 after conversion — 100 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.85 → 3.18 (+0.32) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.
This is not a strict-rule trajectory. It comes from rescue rule S3, which exists because the strict rule returns nothing at all for Distress.
The raw scorer output across the chain is 0.00, 1.09. In the first clip the scorer found no Distress whatsoever (0.00); by the last it is at 1.09.
On the corpus-wide percentile scale those become 0.44, 0.98 — a total move of +0.54.
That is why the strict rule cannot build this chain. Roughly 90 % of the corpus scores exactly zero on Distress, and tied values all collapse onto one point (about 0.45). The clips that genuinely carry Distress sit above 0.85. There is simply nothing in between to step onto, so the only move available is one jump far wider than the 0.25 per-step cap.
What this rule changes: Absent to present. The chain must start at or below 0.05 (the emotion is absent) and end at or above 1.0 (it is clearly present). The most literal reading of 'from not sad to sad'.
What it costs: Says nothing about the SHAPE of the path -- only that it begins absent and ends present. Any intermediate clip is unconstrained.
Same speaker? No similarity score is available here — the emolia clips in this chain are not covered by either speaker-embedding store. The chain therefore rests on the corpus's own speaker/track labelling, which is not the same as a measured check.
Voice consistency: these clips are separate recordings joined together, and no voice-similarity check could be run for this sample, so there is no measurement of how closely the voices match. You may hear the voice shift between segments. Voice conversion has not been applied yet in this build. A planned pass will re-render every segment onto the first segment's voice, which removes this effect entirely.
What was done to this chain. All 2 segments were re-synthesised with ChatterboxVC onto segment 1's voice — including segment 1 itself, converted with itself as the target — and then restored with SIDON. Converting the anchor too is what keeps the room and the reverb the same across the whole chain; leaving it raw put a change of acoustic at the first join. The words, timing and delivery still come from each original clip. The joins are 150–150 ms equal-power crossfades. The chain is normalised as one signal, so the loudness differences between segments are the ones the conversion produced, not a per-clip reset.
Did it unify the voice? On the 250-dimensional Speaker-wavLM-id verification embedding, the worst similarity between any segment and segment 1 was 0.828 before conversion and 0.887 after — it rose by 0.059. Neighbour-to-neighbour the worst pair went 0.828 → 0.887. (The earlier render, with segment 1 left raw, scores 0.739 here.) This chain started at or above 0.80 — already effectively one voice. Across the whole build that is the band where conversion tends to cost identity agreement rather than add it, and a hard identity cut would keep this chain without converting it at all. Judge it by ear against the original above.
Did the emotion survive? Re-scored end to end through the same emotion stack and the same corpus-percentile scale the chain was mined on, Distress moved +0.540 in the original and +0.531 after conversion — 98 % of the delta retained, which is essentially all of it.
Quality. Mean predicted overall quality across the segments went 2.65 → 3.17 (+0.52) on the Empathic-Insight head. The corrected render is 48 kHz because SIDON outputs 48 kHz; the original is the 24 kHz source. Some of what you hear as “cleaner” is that bandwidth, not the conversion.