Gemini Text-to-Speech Skill (1.2x Speed)

Convert text to speech using Google Gemini's TTS API with fixed 1.2x playback speed. Automatically generates standard WAV audio files optimized for video narration.

Features

• Converts text to natural-sounding speech using Gemini 2.5 Pro TTS
• Fixed 1.2x speed for more dynamic video narration
• Outputs standard WAV format (24kHz, 16-bit, mono)
• Supports multiple voice styles
• Handles both English and Chinese text
• Automatic format conversion using ffmpeg

Requirements

• Python 3.x with google-genai package
• ffmpeg (for audio format conversion)
• GOOGLE_API_KEY environment variable (stored in .env file)

Usage

When the user requests text-to-speech conversion, the skill will automatically apply 1.2x speed:

1. Load environment: Source the .env file to get the API key
2. Parse arguments:
   • Text to convert (required) - supports multiple languages
   • --output=filename.wav (optional, default: output.wav)
   • --voice=VoiceName (optional, default: Puck)
   • Speed is automatically set to 1.2x (no need to specify)
3. Generate audio: Run the script to create the WAV file at 1.2x speed
4. Confirm success: Report the output file location and size

Available Voices

• Puck (default) - 中性、清晰 (Neutral, clear)
• Charon - 深沉、权威 (Deep, authoritative)
• Kore - 温暖、友好 (Warm, friendly)
• Fenrir - 强劲、动感 (Strong, dynamic)
• Aoede - 流畅、富有表现力 (Smooth, expressive)

Command Template

The skill now includes its own tts_cli.py script in the skill directory, making it fully portable:

set -a && source .env && set +a && \
source venv/bin/activate && \
python .claude/skills/gemini-tts-fast/tts_cli.py "<text>" --output="<filename>" --voice="<voice>" --speed=1.2

Note: Users need to:

1. Install Python dependencies: pip install google-genai
2. Create a .env file with GOOGLE_API_KEY=your-key
3. Install ffmpeg: brew install ffmpeg (macOS) or equivalent
4. Create a Python virtual environment: python -m venv venv

Error Handling

• If GOOGLE_API_KEY is missing from .env, instruct user to add it
• If ffmpeg is not installed, instruct user to install it (brew install ffmpeg)
• If script fails, show the error message
• If model is unavailable, suggest checking Gemini API status

Examples

Simple usage (English):

/gemini-tts-fast "Hello world"

→ Generates output.wav at 1.2x speed

With custom output:

/gemini-tts-fast "Welcome to our app" --output=welcome.wav

→ Generates welcome.wav at 1.2x speed

With custom voice:

/gemini-tts-fast "Thank you for listening" --output=thanks.wav --voice=Aoede

→ Generates thanks.wav at 1.2x speed with Aoede voice

Chinese text:

/gemini-tts-fast "你好世界" --output=hello_cn.wav --voice=Kore

→ Generates hello_cn.wav at 1.2x speed with Kore voice

Processing script.json:

/gemini-tts-fast @script.json

→ Automatically processes all narration scenes from script.json at 1.2x speed

Technical Details

• Input: Raw PCM data from Gemini API
• Processing: Converts to WAV and applies 1.2x speed using ffmpeg atempo filter
• Output format: RIFF WAVE, 24000 Hz, mono, 16-bit PCM
• Playback speed: Fixed at 1.2x (shortens duration by ~17%)
• Temporary files are automatically cleaned up

Why 1.2x Speed?

1.2x speed is optimal for video narration because:

• Maintains natural speech clarity
• Keeps the content engaging and dynamic
• Reduces video length without sounding rushed
• Standard practice for professional video voiceovers