TTS sentence slices: segmentation, wav output and boundary_lead_ms 70
Sume TTS can return gapless sentence segments. Needs timestamps.words and wav or raw for per-sentence audio_url. A 900-character script costs $0.04275.
Add segmentation {"mode": "sentence"} to a Sume TTS request, with timestamps.words true, and the result returns gapless segments[]. Use a wav or raw container to get a sample-exact audio_url per sentence; a 900-character script is $0.04275.
How the options combine
Rules are from the OpenAPI schema for segmentation.
| Setting | Value | Effect |
|---|---|---|
| mode | sentence (only option) | Cuts at sentence ends |
| boundary_lead_ms | 0 to 500, default 70 | Cut falls this long after the last word; the next segment absorbs the pause |
| emit_audio | default true | With wav or raw: per-segment audio_url. With mp3: timings only |
Request
Per-sentence slices are handy when each sentence maps to one scene.
{
"transcript": "Rain starts at noon. Bring a jacket. Dry by evening.",
"avatar_handle": "@narrator",
"language": "en",
"output_format": {"container": "wav", "sample_rate": 24000, "encoding": "pcm_s16le"},
"timestamps": {"words": true},
"segmentation": {"mode": "sentence", "boundary_lead_ms": 70}
}Gotchas
Segmentation without timestamps.words is invalid. With mp3 you still get segment timings but no slices. Segments are gapless: segment[i].end equals segment[i+1].start, so pauses live at the start of the next sentence.
Sources
Related posts
More in Developers
- Sume TTS sentence segments: boundary_lead_ms 0 vs 500, 12 lines
Ask Sume TTS for words and sentence segmentation and you get gapless wav slices. boundary_lead_ms (default 70) moves each cut 0 to 500 ms after the last word.
- TTS speed 1.2 turns a 30-second read into about 25 seconds
Sume TTS accepts generation_config.speed from 0.6 to 1.5. Speed changes length, not the bill: 450 characters cost $0.021375 at any speed. Test the real length.
- TypeScript: check 8:1 against /v1/images/models before you POST
A 20-line TypeScript guard that reads the aspect_ratio values for a Sume image model and refuses a ratio it does not list, instead of a 400 at request time.
- usage.cost on a Sume video poll is nullable: guard it before you sum
The /v1/videos poll schema allows usage.cost to be null. Sum only completed jobs, keep unknowns apart in a ledger, and compare to the Wan 3.0 $3.75 estimate.
Written by Sume