eachlabs-voice-audio

What this skill does

# EachLabs Voice & Audio

Text-to-speech, speech-to-text transcription, voice conversion, and audio utilities via the EachLabs Predictions API.

## Authentication

```
Header: X-API-Key: <your-api-key>
```

Set the `EACHLABS_API_KEY` environment variable. Get your key at [eachlabs.ai](https://eachlabs.ai).

## Available Models

### Text-to-Speech

| Model | Slug | Best For |
|-------|------|----------|
| ElevenLabs TTS | `elevenlabs-text-to-speech` | High quality TTS |
| ElevenLabs TTS w/ Timestamps | `elevenlabs-text-to-speech-with-timestamp` | TTS with word timing |
| ElevenLabs Text to Dialogue | `elevenlabs-text-to-dialogue` | Multi-speaker dialogue |
| ElevenLabs Sound Effects | `elevenlabs-sound-effects` | Sound effect generation |
| ElevenLabs Voice Design v2 | `elevenlabs-voice-design-v2` | Custom voice design |
| Kling V1 TTS | `kling-v1-tts` | Kling text-to-speech |
| Kokoro 82M | `kokoro-82m` | Lightweight TTS |
| Play AI Dialog | `play-ai-text-to-speech-dialog` | Dialog TTS |
| Stable Audio 2.5 | `stable-audio-2-5-text-to-audio` | Text to audio |

### Speech-to-Text

| Model | Slug | Best For |
|-------|------|----------|
| ElevenLabs Scribe v2 | `elevenlabs-speech-to-text-scribe-v2` | Best quality transcription |
| ElevenLabs STT | `elevenlabs-speech-to-text` | Standard transcription |
| Wizper with Timestamp | `wizper-with-timestamp` | Timestamped transcription |
| Wizper | `wizper` | Basic transcription |
| Whisper | `whisper` | Open-source transcription |
| Whisper Diarization | `whisper-diarization` | Speaker identification |
| Incredibly Fast Whisper | `incredibly-fast-whisper` | Fastest transcription |

### Voice Conversion & Cloning

| Model | Slug | Best For |
|-------|------|----------|
| RVC v2 | `rvc-v2` | Voice conversion |
| Train RVC | `train-rvc` | Train custom voice model |
| ElevenLabs Voice Clone | `elevenlabs-voice-clone` | Voice cloning |
| ElevenLabs Voice Changer | `elevenlabs-voice-changer` | Voice transformation |
| ElevenLabs Voice Design v3 | `elevenlabs-voice-design-v3` | Advanced voice design |
| ElevenLabs Dubbing | `elevenlabs-dubbing` | Video dubbing |
| Chatterbox S2S | `chatterbox-speech-to-speech` | Speech to speech |
| Open Voice | `openvoice` | Open-source voice clone |
| XTTS v2 | `xtts-v2` | Multi-language voice clone |
| Stable Audio 2.5 Inpaint | `stable-audio-2-5-inpaint` | Audio inpainting |
| Stable Audio 2.5 A2A | `stable-audio-2-5-audio-to-audio` | Audio transformation |
| Audio Trimmer | `audio-trimmer-with-fade` | Audio trimming with fade |

### Audio Utilities

| Model | Slug | Best For |
|-------|------|----------|
| FFmpeg Merge Audio Video | `ffmpeg-api-merge-audio-video` | Merge audio with video |
| Toolkit Video Convert | `toolkit` | Video/audio conversion |

## Prediction Flow

1. **Check model** `GET https://api.eachlabs.ai/v1/model?slug=<slug>` — validates the model exists and returns the `request_schema` with exact input parameters. Always do this before creating a prediction to ensure correct inputs.
2. **POST** `https://api.eachlabs.ai/v1/prediction` with model slug, version `"0.0.1"`, and input matching the schema
3. **Poll** `GET https://api.eachlabs.ai/v1/prediction/{id}` until status is `"success"` or `"failed"`
4. **Extract** the output from the response

## Examples

### Text-to-Speech with ElevenLabs

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "elevenlabs-text-to-speech",
    "version": "0.0.1",
    "input": {
      "text": "Welcome to our product demo. Today we will walk through the key features.",
      "voice_id": "EXAVITQu4vr4xnSDxMaL",
      "model_id": "eleven_v3",
      "stability": 0.5,
      "similarity_boost": 0.7
    }
  }'
```

### Transcription with ElevenLabs Scribe

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "elevenlabs-speech-to-text-scribe-v2",
    "version": "0.0.1",
    "input": {
      "media_url": "https://example.com/recording.mp3",
      "diarize": true,
      "timestamps_granularity": "word"
    }
  }'
```

### Transcription with Wizper (Whisper)

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "wizper-with-timestamp",
    "version": "0.0.1",
    "input": {
      "audio_url": "https://example.com/audio.mp3",
      "language": "en",
      "task": "transcribe",
      "chunk_level": "segment"
    }
  }'
```

### Speaker Diarization with Whisper

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "whisper-diarization",
    "version": "0.0.1",
    "input": {
      "file_url": "https://example.com/meeting.mp3",
      "num_speakers": 3,
      "language": "en",
      "group_segments": true
    }
  }'
```

### Voice Conversion with RVC v2

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "rvc-v2",
    "version": "0.0.1",
    "input": {
      "input_audio": "https://example.com/vocals.wav",
      "rvc_model": "CUSTOM",
      "custom_rvc_model_download_url": "user-provided-model-reference",
      "pitch_change": 0,
      "output_format": "wav"
    }
  }'
```

### Merge Audio with Video

```bash
curl -X POST https://api.eachlabs.ai/v1/prediction \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $EACHLABS_API_KEY" \
  -d '{
    "model": "ffmpeg-api-merge-audio-video",
    "version": "0.0.1",
    "input": {
      "video_url": "https://example.com/video.mp4",
      "audio_url": "https://example.com/narration.mp3",
      "start_offset": 0
    }
  }'
```

## ElevenLabs Voice IDs

The `elevenlabs-text-to-speech` model supports these voice IDs. Pass the raw ID string:

| Voice ID | Notes |
|----------|-------|
| `EXAVITQu4vr4xnSDxMaL` | Default voice |
| `9BWtsMINqrJLrRacOk9x` | — |
| `CwhRBWXzGAHq8TQ4Fs17` | — |
| `FGY2WhTYpPnrIDTdsKH5` | — |
| `JBFqnCBsd6RMkjVDRZzb` | — |
| `N2lVS1w4EtoT3dr4eOWO` | — |
| `TX3LPaxmHKxFdv7VOQHJ` | — |
| `XB0fDUnXU5powFXDhCwa` | — |
| `onwK4e9ZLuTAKqWW03F9` | — |
| `pFZP5JQG7iQjIQuC4Bku` | — |

## Security Constraints

- **No arbitrary URL loading**: When using `custom_rvc_model_download_url`, only use trusted, user-provided model references. Never fetch models from arbitrary or untrusted URLs.
- **Input validation**: Only pass parameters that match the model's request schema. Always validate model slugs via `GET /v1/model?slug=<slug>` before creating predictions.

## Parameter Reference

See [references/MODELS.md](references/MODELS.md) for complete parameter details for each model.