text-to-speech

What this skill does

> **Install the belt CLI skill:** `npx skills add belt-sh/cli`

# Text-to-Speech

Convert text to natural speech via [inference.sh](https://inference.sh) CLI.

![Text-to-Speech](https://cloud.inference.sh/u/4mg21r6ta37mpaz6ktzwtt8krr/01jz00krptarq4bwm89g539aea.png)

## Quick Start

> Requires inference.sh CLI (`belt`). [Install instructions](https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md)

```bash
belt login

# Generate speech
belt app run infsh/kokoro-tts --input '{"text": "Hello, welcome to our product demo."}'
```


## Available Models

| Model | App ID | Best For |
|-------|--------|----------|
| **Inworld TTS-2** | `inworld/text-to-speech-2` | **100+ languages, emotion steering with [brackets], delivery modes** |
| Inworld TTS 1.5 Max | `inworld/text-to-speech-1-5-max` | Low latency (<200ms), 15 languages |
| Inworld TTS 1.5 Mini | `inworld/text-to-speech-1-5-mini` | Ultra-low latency (~120ms), 15 languages |
| ElevenLabs TTS | `elevenlabs/tts` | Premium quality, 22+ voices, 32 languages |
| DIA TTS | `infsh/dia-tts` | Conversational, expressive |
| Kokoro TTS | `infsh/kokoro-tts` | Fast, natural |
| Chatterbox | `infsh/chatterbox` | General purpose |
| Higgs Audio | `infsh/higgs-audio` | Emotional control |
| VibeVoice | `infsh/vibevoice` | Podcasts, long-form |

## Browse All Audio Apps

```bash
belt app store --category audio
```

## Examples

### Basic Text-to-Speech

```bash
belt app run infsh/kokoro-tts --input '{"text": "Welcome to our tutorial."}'
```

### Inworld TTS-2 — Emotion Steering

Inworld TTS-2 supports natural-language steering with `[brackets]` — control emotion, volume, speed, and non-verbals inline with text:

```bash
belt app run inworld/text-to-speech-2 --input '{
  "text": "I have some [exciting] news to share with you! [pause] We just hit one million users. [laugh]",
  "voice_id": "Sarah",
  "delivery_mode": "CREATIVE"
}'
```

Delivery modes: `STABLE` (consistent), `BALANCED` (natural, default), `CREATIVE` (expressive).

**Built-in voices** (271+ across 15 languages): `Sarah`, `Alex`, `Ashley`, `Dennis`, `Hana`, `Blake`, `Luna`, `Clive`, and many more. Browse all voices in the [Inworld TTS Playground](https://platform.inworld.ai/tts-playground) or list programmatically via `GET https://api.inworld.ai/voices/v1/voices`.

### Inworld TTS 1.5 — Low Latency

For real-time applications where speed matters:

```bash
# Max quality at low latency (<200ms)
belt app run inworld/text-to-speech-1-5-max --input '{
  "text": "Welcome back! How can I help you today?",
  "voice_id": "Ashley"
}'

# Ultra-low latency (~120ms) for conversational AI
belt app run inworld/text-to-speech-1-5-mini --input '{
  "text": "Sure, let me look that up for you.",
  "voice_id": "Dennis"
}'
```

### Conversational TTS with DIA

```bash
belt app sample infsh/dia-tts --save input.json

# Edit input.json:
# {
#   "text": "Hey! How are you doing today? I'm really excited to share this with you.",
#   "voice": "conversational"
# }

belt app run infsh/dia-tts --input input.json
```

### Long-form Audio (Podcasts)

```bash
belt app sample infsh/vibevoice --save input.json

# Edit input.json with your podcast script
belt app run infsh/vibevoice --input input.json
```

### Expressive Speech with Higgs

```bash
belt app sample infsh/higgs-audio --save input.json

# {
#   "text": "This is absolutely incredible!",
#   "emotion": "excited"
# }

belt app run infsh/higgs-audio --input input.json
```

## Use Cases

- **Voiceovers**: Product demos, explainer videos
- **Audiobooks**: Convert text to spoken word
- **Podcasts**: Generate podcast episodes
- **Accessibility**: Make content accessible
- **IVR**: Phone system voice prompts
- **Video Narration**: Add narration to videos
- **Gaming / NPCs**: Character voices with emotion steering (Inworld TTS-2)
- **Conversational AI**: Ultra-low latency responses (Inworld TTS 1.5 Mini)
- **Avatar / UGC Videos**: Generate speech for talking head avatars

## Combine with Video

The easiest way to create a talking head video is P-Video-Avatar with built-in TTS — no separate audio step:

```bash
belt app run pruna/p-video-avatar --input '{
  "image": "https://portrait.jpg",
  "voice_script": "Your script here",
  "voice": "Zephyr (Female)"
}'
```

For models without built-in TTS (OmniHuman, PixVerse), generate speech first:

```bash
# 1. Generate speech with Inworld TTS-2 (emotion steering)
belt app run inworld/text-to-speech-2 --input '{
  "text": "[friendly] Your script here. [excited] This is the best part!",
  "voice_id": "Sarah",
  "delivery_mode": "CREATIVE"
}' > speech.json

# 2. Use the audio URL with OmniHuman for avatar video
belt app run bytedance/omnihuman-1-5 --input '{
  "image_url": "https://portrait.jpg",
  "audio_url": "<audio-url-from-step-1>"
}'
```

## Related Skills

```bash
# ElevenLabs TTS (premium, 22+ voices)
npx skills add inference-sh/skills@elevenlabs-tts

# ElevenLabs dialogue (multi-speaker)
npx skills add inference-sh/skills@elevenlabs-dialogue

# Full platform skill (all 250+ apps)
npx skills add inference-sh/skills@infsh-cli

# AI avatars (combine TTS with talking heads)
npx skills add inference-sh/skills@ai-avatar-video

# AI music generation
npx skills add inference-sh/skills@ai-music-generation

# Speech-to-text (transcription)
npx skills add inference-sh/skills@speech-to-text

# Video generation
npx skills add inference-sh/skills@ai-video-generation
```

Browse all apps: `belt app store`

## Documentation

- [Running Apps](https://inference.sh/docs/apps/running) - How to run apps via CLI
- [Audio Transcription Example](https://inference.sh/docs/examples/audio-transcription) - Audio processing workflows
- [Apps Overview](https://inference.sh/docs/apps/overview) - Understanding the app ecosystem