Openai Tts Tool

openai-tts-tool

A command-line interface for OpenAI Text-to-Speech API.

Features

• Convert text to natural-sounding speech
• Six high-quality voices with different characteristics
• Two TTS models (tts-1 and tts-1-hd) for speed vs quality
• Multiple audio formats: mp3, opus, aac, flac
• Stream to speakers or save to files
• Adjustable speech speed (0.25 to 4.0)
• Multi-level verbosity logging
• Shell completion support

Installation

# Prerequisites
# - Python 3.14+
# - uv package manager

# Clone and install globally
git clone https://github.com/dnvriend/openai-tts-tool.git
cd openai-tts-tool
uv tool install .

# Verify installation
openai-tts-tool --version

Configuration

Set the OpenAI API key:

export OPENAI_API_KEY="your-api-key-here"

# Add to shell profile for persistence
echo 'export OPENAI_API_KEY="your-api-key-here"' >> ~/.bashrc  # or ~/.zshrc
source ~/.bashrc  # or ~/.zshrc

Usage

# Basic synthesis (stream to speakers)
openai-tts-tool synthesize "Hello, world!"

# Save to file with specific voice
openai-tts-tool synthesize "Hello, world!" --voice nova --output hello.mp3

# High-quality synthesis with custom settings
openai-tts-tool synthesize "Welcome" \
    --voice alloy \
    --model tts-1-hd \
    --output welcome.mp3 \
    --format mp3 \
    --speed 1.2

# Read from stdin
echo "Hello from stdin" | openai-tts-tool synthesize --stdin --output hello.mp3

# List available voices and models
openai-tts-tool list-voices
openai-tts-tool list-models

# Multi-level verbosity for debugging
openai-tts-tool -v synthesize "Test"           # INFO level
openai-tts-tool -vv synthesize "Test"          # DEBUG level
openai-tts-tool -vvv synthesize "Test"         # TRACE with library logs

Options

Synthesize Command

Option	Short	Description
--voice	-v	Voice to use (alloy, echo, fable, onyx, nova, shimmer)
--model	-m	TTS model (tts-1, tts-1-hd)
--output	-o	Output file path (default: stream to speakers)
--format	-f	Audio format (mp3, opus, aac, flac)
--speed	-s	Speech speed (0.25 to 4.0)
--stdin		Read text from stdin

Global Options

Option	Short	Description
--verbose	-v	Verbose output: -v (INFO), -vv (DEBUG), -vvv (TRACE with library logs)
--version		Show version information
--help	-h	Show help message

Voices and Models

Voices

Voice	Description
alloy	Balanced, natural voice (default)
echo	Deeper, authoritative voice
fable	Warm, engaging storyteller
onyx	Deep, confident voice
nova	Clear, professional voice
shimmer	Softer, expressive voice

Models

Model	Quality	Speed	Use Case
tts-1	Standard	Fast	Real-time applications
tts-1-hd	High	Slower	High-quality audio production

Audio Formats

Format	Quality	Use Case
mp3	Good	Maximum compatibility
opus	Excellent	Streaming, low bandwidth
aac	Very Good	Apple devices
flac	Lossless	High-fidelity production

Library Usage

from openai_tts_tool import TTSClient

client = TTSClient(api_key="your-key")

# Stream to speakers
client.synthesize("Hello, world!", voice="alloy")

# Save to file
client.synthesize(
    "Hello, world!",
    voice="nova",
    model="tts-1-hd",
    output="hello.mp3",
    speed=1.2
)

Development

make install          # Install dependencies
make format           # Format code with ruff
make lint             # Run linting with ruff
make typecheck        # Run type checking with mypy
make test             # Run tests with pytest
make security         # Run all security checks
make check            # Run all checks (lint, typecheck, test, security)
make pipeline         # Run full pipeline
make build            # Build package
make run ARGS="..."   # Run openai-tts-tool locally

License

MIT License - see the LICENSE file for details.

Author

Dennis Vriend - GitHub