What it does

gTTS is a Python library and CLI that hits the undocumented speech endpoint inside Google Translate and writes the resulting MP3 to a file, a bytestream, or stdout. It handles the tokenization and text chunking so you can feed it arbitrarily long text without losing intonation on abbreviations or decimals.

The interesting bit

The project survives by reverse-engineering a consumer web feature, not by using Google’s official Cloud Text-to-Speech API. That means no API keys, no billing, and no SLA — just the quiet hope that Google doesn’t shuffle its internal endpoints again.

Key highlights

• CLI and module interfaces; gtts-cli 'hello' --output hello.mp3 is the whole incantation
• Customizable sentence tokenizer for long-text handling without robotic pauses
• Text pre-processor hooks for pronunciation tweaks (e.g., forcing how acronyms sound)
• Outputs to files, file-like objects, or stdout for piping into other tools
• MIT licensed, actively maintained since 2014

Caveats

• Explicitly not affiliated with Google; upstream breaking changes “can occur without notice”
• Uses an undocumented endpoint, so reliability is best-effort
• Different feature set and voice quality from the official Google Cloud Text-to-Speech product

Verdict

Worth a look if you need quick, free TTS for a side project, accessibility script, or audio pipeline prototype. Skip it if you need guaranteed uptime, voice customization, or anything that might end up in production without a fallback plan.