Skip to content

0xsline/OpenChatCut / src/agent/skills/voice

Text-to-Speech (TTS), voiceover, narration placement/sync, and custom sound effects (SFX) generator. Use when the user wants generated speech from text, wants to add/replace/align narration or voiceover for an existing video/timeline, wants to keep existing voiceover synced after visual retiming edits, needs voice audition/selection, or explicitly wants a newly generated/custom sound effect that is not available in the Sound Effects library.

FileKindSize
SKILL.mdskill md17.5 KB
references/minimax-tts.mdmarkdown2.8 KB
references/video-sync.mdmarkdown7.8 KB
references/voices.mdmarkdown12.0 KB

references/voices.md

View on GitHub
# Voice Presets

This page is the curated preset snapshot the agent works from. The backend
holds the source of truth and keeps this file in sync.

Use tags as selection filters. Use descriptions as practical hints. Age,
accent, pronunciation, and exact duration are not stable controls unless the
TTS provider exposes explicit parameters for them.

This curated snapshot covers Doubao, ElevenLabs, and MiniMax only. Inworld,
Fish Audio, Speechify, OpenAI, Gemini, Mistral, and Cartesia have no bundled
OpenChatCut preset/sample catalog. For those providers, require a concrete
provider-specific `voiceId` from the user or their provider account; never
invent a preset or `/voice-samples/...` URL.

## Doubao Chinese Voices

Use Doubao for Chinese-optimized narration.

| Preset               | Official display | English display      | Tags                                                           | Selection hint                                                |
| -------------------- | ---------------- | -------------------- | -------------------------------------------------------------- | ------------------------------------------------------------- |
| `vivi`               | Vivi             | Vivi                 | female, young, friendly, general, short-explainer              | Everyday Chinese narration, product intros, short explainers  |
| `xiaohe`             | 小何             | Xiaohe               | female, young, soft, calm, tutorial, walkthrough               | Calm tutorials, product walkthroughs, instructional narration |
| `yunzhou`            | 云舟             | Yunzhou              | male, young, neutral, business, explainer                      | Business explainers, factual reads, product narration         |
| `xiaotian`           | 小天             | Xiaotian             | male, young, bright, upbeat, casual                            | Casual creator-style videos and light social narration        |
| `naiqimengwa`        | 奶气萌娃         | Childlike Boy        | male, child, cute, storybook, character                        | Cute boy character lines and children's stories               |
| `yingtaowanzi`       | 樱桃丸子         | Cherry Voice         | female, child, cartoon, roleplay, character                    | Animated/kid-oriented character dialogue                      |
| `wenroumama`         | 温柔妈妈         | Warm Mom             | female, middle-aged, warm, family, gentle                      | Family, lifestyle, parenting, gentle explanations             |
| `zhixingnv`          | 知性女声         | Knowledgeable Female | female, middle-aged, calm, knowledge, explainer                | Education, culture, thoughtful explainers                     |
| `dayi`               | 大壹             | Dayi                 | male, young, steady, formal, documentary, video-voiceover      | Formal voiceover, documentary, corporate narration            |
| `jitangnv`           | 鸡汤女           | Inspirational Female | female, young, warm, inspirational, emotional, video-voiceover | Motivational, uplifting, emotional video narration            |
| `liuchang`           | 流畅女声         | Smooth Female        | female, young, smooth, polished, video-voiceover, narration    | Clean product narration and polished video voiceover          |
| `ruyayichen`         | 儒雅逸辰         | Yichen               | male, young, elegant, premium, culture, video-voiceover        | Cultural, premium, poetic, documentary-style narration        |
| `morgan`             | Morgan           | Morgan               | male, middle-aged, deep, knowledge, explainer                  | Deep knowledge explainer, documentary, serious narration      |
| `qingcang`           | 擎苍             | Qingcang             | male, old-like, authoritative, audiobook, character            | Weighty narration, dramatic reads, audiobook scenes           |
| `huiben`             | 儿童绘本         | Storybook Voice      | female, young, gentle, storybook, audiobook                    | Gentle storybook or bedtime-style narration                   |
| `popo`               | 婆婆             | Grandma              | female, old, warm, story, character                            | Grandmother characters, folk stories, nostalgic narration     |
| `yuanboxiaoshu`      | 渊博小叔         | Erudite Uncle        | male, middle-aged, knowledge, calm, explainer                  | Calm explainers, cultural commentary, educational narration   |
| `baqiqingshu`        | 霸气青叔         | Confident Uncle      | male, middle-aged, confident, audiobook, narrative             | Audiobook narration, long-form stories, dramatic reads        |
| `shuanglangshaonian` | 爽朗少年         | Cheerful Teen        | male, young, cheerful, youthful, roleplay, character           | Bright roleplay, teen/creator dialogue, youthful scenes       |
| `tangseng`           | 唐僧             | Tang Seng            | male, old-like, calm, roleplay, character                      | Monk-like dialogue, traditional stories, steady narration     |

Important caveat: the current Doubao `seed-audio-1.0` endpoint does not expose
emotion, performance-prompt, or dialect controls. `qingcang` and `tangseng` are
old-like approximations, not guaranteed old-male controls.

### Doubao Control Support

Base controls for curated Doubao voices: `speedRatio`, `loudnessRatio`, and
native `pitch` (`pitch_rate` −12–12). `text_prompt` carries the plain narration
text; the speaker comes from `references[].speaker`.

A raw Volcengine speaker ID the user supplies (for example `S_5wulnA7i2`, or an
official `zh_female_..._bigtts` ID) is a valid `voiceId` and is sent verbatim as
`references[].speaker`; it takes priority over curated aliases. When Settings →
配音 / TTS → 豆包 TTS has a default 音色 ID, omitting `voiceId` uses that saved
speaker. Never shorten, translate, or alter a supplied speaker ID.

## ElevenLabs English / Multilingual Voices

Use ElevenLabs for English or multilingual narration. Official accent labels
are English-source voice cues, not target-language accent controls.

| Preset      | Tags                                                       | Selection hint                                             |
| ----------- | ---------------------------------------------------------- | ---------------------------------------------------------- |
| `amelia`    | female, young, upbeat, narrative-story, social-media       | Story reads, podcast intros, reels, enthusiastic ads       |
| `brittney`  | female, young, upbeat, social-media, fun                   | Creator videos, recaps, hot-topic commentary, how-to clips |
| `hope`      | female, young, upbeat, clear, social-media                 | Crisp short-form narration and quick explainers            |
| `jessica`   | female, middle-aged, calm, conversational, narrative-story | Composed narration, confident product copy, direct reads   |
| `arabella`  | female, young, gentle, emotive, narrative-story            | Fantasy, romance, wellness, atmospheric stories            |
| `jane`      | female, old, professional, audiobook, narrative-story      | Long-form book pacing and classic narration                |
| `maria`     | female, old, calm, grandmother, narrative-story            | Grandmother-style narration and reflective story delivery  |
| `mark`      | male, young, casual, conversational, natural               | Dialogue, assistant-style replies, informal scripts        |
| `frederick` | male, middle-aged, calm, documentary, narrative-story      | History, science, mystery, factual films                   |
| `peter`     | male, middle-aged, confident, credible, narrative-story    | Trustworthy narration, explainers, brand reads             |
| `james`     | male, middle-aged, deep, husky, narrative-story            | Audiobooks, heavier story narration, professional VO       |
| `jon`       | male, middle-aged, calm, grounded, narrative-story         | Clear messaging, commercials, trustworthy narration        |
| `sully`     | male, old, deep, storyteller, narrative-story              | Deep elderly narration and warm authoritative reads        |
| `david`     | male, old, deep, storyteller, narrative-story              | Classic audiobook passages and grounded dramatic reads     |
| `alex`      | male, young, confident, entertainment-tv, social-media     | YouTube, shorts, entertainment clips                       |

### ElevenLabs Control Support

All current curated ElevenLabs presets support the same request-level controls:
`modelId`, `speed`, and `stability`.

Important limits:

- These controls are provider/model-level controls, not per-voice capability
  switches. No current curated ElevenLabs preset is excluded from them, but the
  audible result varies by the source voice's natural delivery.
- For ElevenLabs `eleven_v3`, inline audio tags are available when the user
  asks for expressive delivery such as emotion, tone, nonverbal cues, accent
  hints, or local pacing. Official examples fit these useful TTS categories:
  emotion/tone tags such as `[happy]`, `[sad]`, `[angry]`, `[excited]`,
  `[curious]`, `[sarcastic]`, `[crying]`, `[annoyed]`, `[appalled]`,
  `[thoughtful]`, `[surprised]`, and `[mischievously]`; vocal delivery and
  nonverbal cue tags such as `[whispers]`, `[laughs]`, `[sighs]`, `[exhales]`,
  `[inhales deeply]`, `[clears throat]`, `[snorts]`, `[swallows]`,
  `[wheezing]`, and `[coughs]`;
  pacing/pause/local speed tags such as `[slowly]`, `[pause]`,
  `[short pause]`, `[long pause]`, `[rushed]`, and `[drawn out]`; and
  accent/special-performance tags such as
  `[strong X accent]`, for example `[strong French accent]`, plus `[sings]`,
  `[singing]`, `[woo]`, and `[pirate voice]`. Official examples are
  non-exhaustive; similar auditory tags can be tried when the user explicitly
  asks for that delivery and the tag describes how the voice should sound, not
  a visual action. Write tags directly in `text`, close to the short phrase
  they should affect. Treat tags as local guidance, not paragraph-wide controls.
- For pauses and pacing in `eleven_v3`, use punctuation, text structure,
  shorter generated segments, or local audio tags such as `[short pause]` and
  `[slowly]` when needed.

## MiniMax Chinese Voices

Use MiniMax when that vendor is configured and the user wants MiniMax TTS
(or Doubao/ElevenLabs are off). Pass `provider: "minimax"` and a system
`voiceId` (or a cloned voice id from the user's MiniMax account).

| Preset | Tags | Selection hint |
| --- | --- | --- |
| `female-yujie` | female, adult, clear, default, narration | Default MiniMax system voice; product / general narration |
| `female-shaonv` | female, young, bright, casual | Youthful / light social reads |
| `female-chengshu` | female, mature, calm, professional | Mature professional narration |
| `female-tianmei` | female, sweet, soft, lifestyle | Softer lifestyle / gentle product lines |
| `male-qn-qingse` | male, young, clear, neutral | Everyday male Chinese narration |
| `male-qn-jingying` | male, adult, steady, business | Business / confident explainer tone |

See [minimax-tts.md](minimax-tts.md) for tool knobs (`speed`, `emotion`).

### MiniMax Control Support

- `speed` roughly `0.5–2` (server enforces range).
- Optional `emotion` labels when the model accepts them: happy, sad, angry,
  fearful, disgusted, surprised, calm, fluent, whisper.
- Do **not** pass Doubao-only fields (`speedRatio`, `loudnessRatio`, …) or
  ElevenLabs-only fields (`modelId`, `stability`).

## Audition Samples

Each curated preset has a local editor sample under:

```text
/voice-samples/<provider>-<preset>.mp3
```

Examples:

- `/voice-samples/doubao-morgan.mp3`
- `/voice-samples/doubao-zhixingnv.mp3`
- `/voice-samples/elevenlabs-peter.mp3`
- `/voice-samples/elevenlabs-sully.mp3`

Prefer these real samples when rendering a curated voice audition widget before
calling `submit_voice`. MiniMax system voices may not ship local samples; offer
only cataloged names/tags and confirm the exact ID before `submit_voice`. Never
derive a sample path for a provider or voice absent from this file.
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 472kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall