BACON-AI TTS

FREE Text-to-Speech for Claude Code using Microsoft Edge TTS

Warning This project is currently in Alpha testing. It is functional and actively used in development, but you may encounter bugs, rough edges, or breaking changes between releases. Feedback, bug reports, and contributions are very welcome! Please open an issue if you run into problems.

Meet Your AI Project Team. They Talk.

Picture this: You're running three AI coding projects across two machines. You step away to make coffee. From across the room, a proper British accent calls out:

"Darling, I'm waiting for you. The test results are in and we need to discuss the deployment strategy."

That's Elisabeth (en-GB-SoniaNeural) - your Operations Coordinator. She announces every tool execution before it happens ("Hello, I'm about to run the test suite"), delivers the daily project debrief, reports on SSC scores, lessons learned, and the self-annealing mitigations the team installed overnight to prevent yesterday's bugs from ever happening again. If you ignore her approval request for more than 30 seconds, she gets... persistent.

You sit down. A cheerful Norwegian accent pipes up:

"Wakey wakey! We need your input here. The team achieved 47 tasks overnight but three need your approval."

That's Finn (nb-NO-FinnNeural) - your Innovation Lead. Every time a tool completes successfully, Finn confirms it. "Hei! Fantastisk! The build passed!" He summarises what the team achieved and isn't afraid to tell you when something went wrong either.

Then Katja (de-DE-KatjaNeural), your Research Specialist, cuts in with her precise German tone to explain a notification about the authentication module. If a tool request sits unapproved too long, Katja will nudge you: "Excuse me, but there is a tool request waiting for your approval." Very German. Very efficient.

Giuseppe (it-IT-GiuseppeNeural) follows up with the research details of the integration failures - expressive, thorough, Italian. George (en-GB-RyanNeural), the Systems Architect, gives you the strategic debrief of today's architecture decisions. Lily (en-GB-LibbyNeural) acknowledges every prompt you submit: "Brilliant! Let's build something amazing." Connor (en-IE-ConnorNeural), the Irish DevOps Engineer, handles the operational alerts and pre-compaction warnings. And Daniel (en-GB-ThomasNeural) runs QA inspection at shutdown.

This is not science fiction. This is how we actually work. Every day.

How It Actually Works

The voice system is hook-driven - deeply integrated into Claude Code's event lifecycle:

flowchart LR
    A["Claude Code Event"] --> B["Hook Trigger"]
    B --> C{"Voice Mode?"}
    C -->|Full| D["Select Persona"]
    C -->|Tool-only| E["Brief Name Only"]
    C -->|Beep| F["Audio Cue"]
    C -->|Silent| G["Nothing"]

    D --> H["Elisabeth: PreToolUse"]
    D --> I["Finn: PostToolUse Success"]
    D --> J["Katja: Notifications"]
    D --> K["George: SubagentStop"]
    D --> L["Lily: UserPromptSubmit"]
    D --> M["Connor: PreCompact/Ops"]

    H & I & J & K & L & M --> N["Edge TTS"] --> O["Audio Playback"]

    style A fill:#e94560,stroke:#fff,color:#fff
    style D fill:#0f3460,stroke:#fff,color:#fff
    style N fill:#16c79a,stroke:#fff,color:#fff

Each persona is triggered by a specific Claude Code hook event - Elisabeth announces before every tool, Finn celebrates after success, Katja alerts on notifications, George reports on sub-agent completions. It's not random chatter - it's a structured information architecture delivered through voice.

The International Voice Team

Persona	Voice	Country	Hook Trigger	What They Do
Elisabeth	en-GB-SoniaNeural	UK	PreToolUse	"I'm about to execute..." - announces every action before it happens
Lily	en-GB-LibbyNeural	UK	UserPromptSubmit	"Brilliant! Let's build..." - acknowledges your prompts
Finn	nb-NO-FinnNeural	Norway	PostToolUse	"Fantastisk! Tests passed!" - celebrates completions, summarises progress
George	en-GB-RyanNeural	UK	SubagentStop	"The architecture review is complete..." - reports on sub-agent results
Katja	de-DE-KatjaNeural	Germany	Notifications	"Excuse me, there is a request waiting..." - alerts and nudges
Connor	en-IE-ConnorNeural	Ireland	PreCompact/Ops	"Context window approaching limit..." - operational warnings
Daniel	en-GB-ThomasNeural	UK	Stop/QA	"QA inspection complete..." - quality checks at shutdown
Giuseppe	it-IT-GiuseppeNeural	Italy	Research	"The analysis reveals..." - detailed research findings

Multi-Agent Teams with Voice

When you run BACON-AI Agent Teams with up to 10 specialized agents working in parallel, each agent gets their own voice persona. The team lead (Elisabeth) orchestrates in delegate mode - she never writes code, only coordinates. The frontend developer, backend developer, tester, and architect each report progress in their own international accent.

You don't stare at terminal logs. You listen to your team like a real project manager. Walk around, make lunch, pet the dog. Your AI team keeps you posted across multiple machines.

When It Gets Too Much

And when the constant international jabber drives you absolutely crazy? One command to switch modes:

python3 voice_config.py beep    # Subtle audio cues only
python3 voice_config.py silent  # Blissful silence
python3 voice_config.py full    # Bring back the team

Mode	What You Hear
Full	Elisabeth announces, Finn celebrates, Katja nudges, Giuseppe explains, George debriefs
Tool-only	Brief tool name announcements - just the essentials
Beep	Quick audio cues - boop task started, ding task done
Silent	Nothing. Just you and the code. The team works quietly

It's like being back in an international project team. Except these colleagues never take sick days, never argue about the thermostat, and actually read the documentation. And if you tell them to shut up, they don't take it personally.

What is BACON-AI TTS?

BACON-AI TTS brings unlimited, free, natural-sounding text-to-speech to Claude Code and other MCP-compatible AI tools. No API keys, no subscriptions, no limits — just high-quality voice synthesis powered by Microsoft Edge's TTS engine. 400+ voices across 100+ languages, completely free, forever.

✨ Key Features

Feature	Description
100% Free	Zero cost. No API keys. Unlimited usage.
400+ Voices	Natural-sounding voices across 100+ languages
MCP Native	First-class integration with Claude Code
Zero Setup	No accounts, no configuration files, just install
Multi-Format	MP3, WAV, and more output formats
Voice Search	Filter by language, gender, style, and region
Auto-Play	Generated audio plays immediately
Batch Support	Process long text in chunks
SSML Ready	Fine-grained control over speech synthesis

🎭 Popular Voices

Language	Voice Name	Gender	Description
🇬🇧 English (UK)	`en-GB-SoniaNeural`	Female	British accent, professional
🇺🇸 English (US)	`en-US-AriaNeural`	Female	American accent, clear
🇺🇸 English (US)	`en-US-GuyNeural`	Male	American accent, friendly
🇳🇴 Norwegian	`nb-NO-FinnNeural`	Male	Norwegian accent, warm
🇩🇪 German	`de-DE-KatjaNeural`	Female	German accent, professional
🇫🇷 French	`fr-FR-DeniseNeural`	Female	French accent, elegant
🇪🇸 Spanish	`es-ES-ElviraNeural`	Female	Spanish accent, expressive
🇯🇵 Japanese	`ja-JP-NanamiNeural`	Female	Japanese accent, polite
🇨🇳 Chinese	`zh-CN-XiaoxiaoNeural`	Female	Mandarin, natural
🇮🇳 Hindi	`hi-IN-SwaraNeural`	Female	Hindi accent, clear

See full voice list →

🚀 How It Works

flowchart LR
    A[Claude Code] -->|MCP Protocol| B[Edge TTS Server]
    B -->|Text Input| C[Microsoft Edge TTS API]
    C -->|Audio Stream| D[Audio File]
    D -->|Auto-play| E[🔊 User Hears Speech]

    style A fill:#2563eb,stroke:#1e40af,color:#fff
    style B fill:#7c3aed,stroke:#6d28d9,color:#fff
    style C fill:#059669,stroke:#047857,color:#fff
    style D fill:#dc2626,stroke:#b91c1c,color:#fff
    style E fill:#ea580c,stroke:#c2410c,color:#fff

🏗️ Architecture

graph TB
    subgraph "MCP Client (Claude Code)"
        A[User Task] --> B[Claude AI]
        B --> C[MCP Protocol]
    end

    subgraph "Edge TTS Server"
        C --> D[MCP Handler]
        D --> E[Voice Registry]
        D --> F[TTS Engine]
        F --> G[Audio Player]
    end

    subgraph "External Services"
        F --> H[Microsoft Edge TTS API]
    end

    G --> I[🔊 Audio Output]

    style A fill:#f3f4f6,stroke:#9ca3af
    style B fill:#2563eb,stroke:#1e40af,color:#fff
    style C fill:#7c3aed,stroke:#6d28d9,color:#fff
    style D fill:#059669,stroke:#047857,color:#fff
    style E fill:#dc2626,stroke:#b91c1c,color:#fff
    style F fill:#ea580c,stroke:#c2410c,color:#fff
    style G fill:#8b5cf6,stroke:#7c3aed,color:#fff
    style H fill:#10b981,stroke:#059669,color:#fff
    style I fill:#f59e0b,stroke:#d97706,color:#fff

🎤 Voice Selection Flow

flowchart TD
    A[Start] --> B{Know exact voice?}
    B -->|Yes| C[Use voice name directly]
    B -->|No| D[Search by language]
    D --> E[Filter by gender]
    E --> F[Filter by style]
    F --> G[Preview voice]
    G --> H{Satisfied?}
    H -->|No| D
    H -->|Yes| I[Select voice]
    C --> J[Generate speech]
    I --> J
    J --> K[🔊 Audio output]

    style A fill:#f3f4f6,stroke:#9ca3af
    style B fill:#fef3c7,stroke:#fbbf24
    style C fill:#2563eb,stroke:#1e40af,color:#fff
    style D fill:#7c3aed,stroke:#6d28d9,color:#fff
    style E fill:#059669,stroke:#047857,color:#fff
    style F fill:#dc2626,stroke:#b91c1c,color:#fff
    style G fill:#ea580c,stroke:#c2410c,color:#fff
    style H fill:#fef3c7,stroke:#fbbf24
    style I fill:#8b5cf6,stroke:#7c3aed,color:#fff
    style J fill:#10b981,stroke:#059669,color:#fff
    style K fill:#f59e0b,stroke:#d97706,color:#fff

🔄 Integration Pattern

sequenceDiagram
    participant U as User
    participant C as Claude Code
    participant M as MCP Server
    participant E as Edge TTS API

    U->>C: Give task
    C->>C: Work on task
    C->>M: text_to_speech("Starting build", voice)
    M->>E: Convert text to speech
    E-->>M: Audio stream
    M->>M: Save & auto-play
    M-->>C: Audio file path
    U->>U: 🔊 Hears announcement
    C->>C: Continue working
    C->>M: text_to_speech("Build complete!", voice)
    M->>E: Convert text to speech
    E-->>M: Audio stream
    M->>M: Save & auto-play
    M-->>C: Audio file path
    U->>U: 🔊 Hears completion
    C-->>U: Task complete

📦 Installation

Prerequisites

Node.js 18+ or Python 3.10+
Claude Code (or any MCP-compatible client)

Install via npm (Node.js)

npm install -g bacon-ai-tts

Install via pip (Python)

pip install bacon-ai-tts

⚙️ Configuration

Claude Code Setup

Add to your Claude Code MCP settings (~/.claude/mcp.json or %USERPROFILE%\.claude\mcp.json):

{
  "mcpServers": {
    "edge-tts": {
      "command": "bacon-ai-tts",
      "args": [],
      "env": {}
    }
  }
}

That's it! No API keys, no credentials, no configuration files.

Alternative: Run from source

git clone https://github.com/bacon-ai/bacon-ai-tts.git
cd bacon-ai-tts
npm install  # or: pip install -e .
npm start    # or: python -m edge_tts_mcp

🎯 Usage Examples

Basic Text-to-Speech

# From Claude Code (Claude calls this automatically)
edge-tts:text_to_speech(
    text="Hello, starting the build process now.",
    voice_name="en-GB-SoniaNeural"
)

Search for Voices

# Find all Norwegian voices
edge-tts:search_voices(
    language="nb-NO"
)

# Find female American voices
edge-tts:search_voices(
    language="en-US",
    gender="Female"
)

# Find voices with specific style
edge-tts:search_voices(
    language="en-US",
    style="cheerful"
)

Batch Processing Long Text

# Automatically splits long text into chunks
edge-tts:text_to_speech(
    text="Very long text content here... (continues for many paragraphs)",
    voice_name="en-US-AriaNeural",
    batch_mode=True
)

SSML for Advanced Control

# Fine-grained speech control
edge-tts:text_to_speech(
    text="""
    <speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis">
        <prosody rate="slow" pitch="+2st">
            Build completed successfully!
        </prosody>
        <break time="500ms"/>
        <prosody rate="fast">
            All tests passed.
        </prosody>
    </speak>
    """,
    voice_name="en-US-GuyNeural",
    ssml=True
)

🎬 Real-World Use Cases

1. Build Notifications

# Claude announces build stages
"Starting TypeScript compilation..."  # 🔊 Elisabeth (UK)
"Running test suite..."               # 🔊 Elisabeth (UK)
"Build complete! All tests passed."   # 🔊 Finn (Norwegian)

2. Accessibility Workflow

# Screen reader for code review
"Reviewing file: main.py"
"Found 3 issues: Line 42, unused variable..."
"Suggesting fix: Remove unused import."

3. Multi-Language Development

# Team notifications in native languages
"Compilation terminée!"        # 🔊 French
"Alle Tests bestanden!"        # 🔊 German
"ビルド完了しました！"          # 🔊 Japanese

4. Long-Running Task Updates

# Periodic status updates
"Database migration: 25% complete..."  # Every 30 seconds
"Database migration: 50% complete..."
"Database migration: Complete!"

5. Error Announcements

# Critical alerts
"WARNING: Syntax error detected in main.py"
"ERROR: Build failed. Check logs for details."

🗂️ Voice Catalog

📋 View Full Voice List (400+ voices)

English Voices

Locale	Voice Name	Gender	Style Tags
en-US	`en-US-AriaNeural`	Female	chat, customerservice, narration-professional, newscast-casual
en-US	`en-US-GuyNeural`	Male	newscast, angry, cheerful, sad, excited
en-US	`en-US-JennyNeural`	Female	assistant, chat, customerservice, newscast
en-GB	`en-GB-SoniaNeural`	Female	cheerful, sad, professional
en-GB	`en-GB-RyanNeural`	Male	chat, cheerful
en-AU	`en-AU-NatashaNeural`	Female	-
en-CA	`en-CA-ClaraNeural`	Female	-
en-IN	`en-IN-NeerjaNeural`	Female	-

European Languages

Locale	Voice Name	Gender
de-DE	`de-DE-KatjaNeural`	Female
fr-FR	`fr-FR-DeniseNeural`	Female
es-ES	`es-ES-ElviraNeural`	Female
it-IT	`it-IT-ElsaNeural`	Female
pt-PT	`pt-PT-RaquelNeural`	Female
nl-NL	`nl-NL-ColetteNeural`	Female
nb-NO	`nb-NO-FinnNeural`	Male
sv-SE	`sv-SE-SofieNeural`	Female
da-DK	`da-DK-ChristelNeural`	Female

Asian Languages

Locale	Voice Name	Gender
ja-JP	`ja-JP-NanamiNeural`	Female
zh-CN	`zh-CN-XiaoxiaoNeural`	Female
ko-KR	`ko-KR-SunHiNeural`	Female
hi-IN	`hi-IN-SwaraNeural`	Female
th-TH	`th-TH-PremwadeeNeural`	Female

Full list available in documentation →

🛠️ MCP Tools Reference

`text_to_speech`

Generate speech from text.

Parameters:

text (string, required): Text to convert to speech
voice_name (string, optional): Voice identifier (default: en-GB-SoniaNeural)
output_format (string, optional): mp3, wav (default: mp3)
rate (string, optional): Speech rate: x-slow, slow, medium, fast, x-fast
pitch (string, optional): Pitch adjustment: -20% to +20%
volume (string, optional): Volume: silent, x-soft, soft, medium, loud, x-loud
auto_play (boolean, optional): Auto-play after generation (default: true)
ssml (boolean, optional): Input is SSML markup (default: false)

Returns:

{
  "status": "success",
  "audio_file": "/tmp/edge-tts-12345.mp3",
  "duration_seconds": 3.2,
  "voice_used": "en-GB-SoniaNeural",
  "played": true
}

`search_voices`

Search and filter available voices.

Parameters:

language (string, optional): Language code (e.g., en-US, ja-JP)
gender (string, optional): Male, Female
style (string, optional): Voice style tag (e.g., cheerful, professional)
region (string, optional): Region code (e.g., US, GB, AU)

Returns:

{
  "status": "success",
  "voices": [
    {
      "name": "en-GB-SoniaNeural",
      "language": "en-GB",
      "gender": "Female",
      "styles": ["cheerful", "sad", "professional"]
    }
  ],
  "count": 12
}

`list_voices`

List all available voices (400+).

Parameters: None

Returns:

{
  "status": "success",
  "voices": [...],
  "count": 400
}

`check_status`

Check server health and voice availability.

Returns:

{
  "status": "healthy",
  "voices_available": 400,
  "edge_tts_version": "6.1.9",
  "server_uptime": "2h 34m"
}

🎨 BACON-AI Integration

This server is part of the BACON-AI ecosystem — a suite of AI development tools and MCP servers for Claude Code.

Related Projects

bacon-ai-voice - Full STT+TTS voice control for Claude Code
bacon-mesh - Multi-machine AI agent coordination
mcp-linear - Linear issue tracking integration
mcp-playwright - Browser automation for Claude

Voice Team Personalities

BACON-AI uses character-based voice assignments:

Character	Voice	Role	Locale
Elisabeth	`en-GB-SoniaNeural`	Technical announcements	British 🇬🇧
Finn	`nb-NO-FinnNeural`	Task confirmations	Norwegian 🇳🇴
Aoede	`en-GB-AbbiNeural`	Debugging humor	Scottish 🏴󠁧󠁢󠁳󠁣󠁴󠁿
Fenrir	`nb-NO-PernilleNeural`	Solutions delivery	Norwegian 🇳🇴

📚 Documentation

🤝 Contributing

Contributions welcome! Please read our Contributing Guide first.

Development Setup

git clone https://github.com/bacon-ai/bacon-ai-tts.git
cd bacon-ai-tts
npm install  # or: pip install -e ".[dev]"
npm run dev  # or: python -m edge_tts_mcp --dev

Running Tests

npm test       # or: pytest
npm run lint   # or: ruff check .

📄 License

MIT License - see LICENSE file for details.

🐛 Bug Reports & Feature Requests

Issues: GitHub Issues
Discussions: GitHub Discussions

💬 About BACON-AI

BACON-AI is an open-source initiative building advanced AI development tools, with a focus on:

Multi-agent coordination
Voice-driven workflows
Cross-platform development
MCP server ecosystem

Contact

Created by: BACON-AI Team
Email: hello@bacon-ai.cloud
GitHub: @BACON-AI-CLOUD
Website: bacon-ai.cloud

🌟 Support This Project

If you find this useful:

⭐ Star the repo
🐛 Report bugs
💡 Suggest features
🔀 Submit PRs
📢 Share with your team

Made with ❤️ by the BACON-AI team
_{Powered by Microsoft Edge TTS • Built for Claude Code • MCP Native}

What is this? • Installation • Usage • Docs • Contributing

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
assets		assets
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

BACON-AI TTS

Meet Your AI Project Team. They Talk.

How It Actually Works

The International Voice Team

Multi-Agent Teams with Voice

When It Gets Too Much

What is BACON-AI TTS?

✨ Key Features

🎭 Popular Voices

🚀 How It Works

🏗️ Architecture

🎤 Voice Selection Flow

🔄 Integration Pattern

📦 Installation

Prerequisites

Install via npm (Node.js)

Install via pip (Python)

⚙️ Configuration

Claude Code Setup

Alternative: Run from source

🎯 Usage Examples

Basic Text-to-Speech

Search for Voices

Batch Processing Long Text

SSML for Advanced Control

🎬 Real-World Use Cases

1. Build Notifications

2. Accessibility Workflow

3. Multi-Language Development

4. Long-Running Task Updates

5. Error Announcements

🗂️ Voice Catalog

English Voices

European Languages

Asian Languages

🛠️ MCP Tools Reference

text_to_speech

search_voices

list_voices

check_status

🎨 BACON-AI Integration

Related Projects

Voice Team Personalities

📚 Documentation

🤝 Contributing

Development Setup

Running Tests

📄 License

🐛 Bug Reports & Feature Requests

💬 About BACON-AI

Contact

🌟 Support This Project

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

`text_to_speech`

`search_voices`

`list_voices`

`check_status`

Packages