Microsoft Edge's speech engine, no browser or API key required

What it does The library acts as a remote client for Microsoft Edge’s online text-to-speech service, letting you synthesize speech and generate SRT subtitle files from plain text. It exposes both a Python module and command-line tools that can list available neural voices, adjust prosody, and write audio to MP3. You do not need to run Windows, install the Edge browser, or obtain an API key to use it.

The interesting bit The project is essentially well-behaved glue code for an endpoint Microsoft hosts for its browser, yet it has remained stable enough to accumulate over eleven thousand stars. The maintainer removed custom SSML support after discovering the service aggressively validates markup, rejecting anything that Edge itself could not have produced.

Key highlights

• Large catalog of neural voices spanning dozens of languages and locales
• Adjustable rate, volume, and pitch via command-line flags or Python arguments
• Generates MP3 audio and matching SRT subtitle files in one pass
• edge-playback helper for immediate previewing (relies on mpv except on Windows)
• Adopted by downstream projects including Home Assistant integrations and podcast generators

Caveats

• Custom SSML is unsupported because the service restricts markup to a single <voice> tag containing one <prosody> tag
• edge-playback requires the mpv player on non-Windows systems
• Negative values for rate, volume, or pitch require an equals sign to avoid being misinterpreted by the argument parser

Verdict Developers who need quick, free, multilingual neural speech synthesis for scripts or automation will find this a pragmatic shortcut. Anyone requiring custom SSML or offline operation should look elsewhere, since the tool is strictly a client for Microsoft’s online service.