#4949th of 93 in VoiceFor voice, speech and music

realtimetts

Python library that streams LLM text into low-latency speech across many TTS engines

Stars
4.1k
License
MIT
Last commit
Nov 2026
Last release
Sep 2026
Language
Python
Stars, last 29 weeks+202
25 Apr 2026: 3,859 stars2 May 2026: 3,868 stars9 May 2026: 3,883 stars16 May 2026: 3,899 stars23 May 2026: 3,912 stars30 May 2026: 3,920 stars6 Jun 2026: 3,930 stars13 Jun 2026: 3,937 stars20 Jun 2026: 3,948 stars27 Jun 2026: 3,958 stars4 Jul 2026: 3,966 stars11 Jul 2026: 3,971 stars18 Jul 2026: 3,978 stars25 Jul 2026: 3,989 stars1 Aug 2026: 3,994 stars8 Aug 2026: 3,997 stars15 Aug 2026: 4,003 stars22 Aug 2026: 4,006 stars29 Aug 2026: 4,013 stars5 Sept 2026: 4,017 stars12 Sept 2026: 4,021 stars19 Sept 2026: 4,028 stars26 Sept 2026: 4,032 stars3 Oct 2026: 4,037 stars10 Oct 2026: 4,043 stars17 Oct 2026: 4,044 stars24 Oct 2026: 4,046 stars31 Oct 2026: 4,050 stars7 Nov 2026: 4,055 stars14 Nov 2026: 4,061 stars

3,859 on 25 Apr 20264,061 on 14 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

RealtimeTTS is a Python text-to-speech library that takes strings, generators or LLM token streams and plays audio as sentences arrive, writes WAV files, or hands chunks to another process. It wraps about 30 engines (system voices, cloud APIs, local neural models) with fallback between them. The recommended QwenEngine runs Qwen3-TTS natively on an NVIDIA GPU or CPU, and can also run as an OpenAI-compatible HTTP server.

Who it is for: Developers adding streaming speech output to LLM or voice-agent apps

Strengths

  • One TextToAudioStream API over about 30 engines, local and cloud, with fallback engines
  • feed() accepts generators, so speech starts while an LLM is still streaming
  • Qwen server exposes /v1/audio/speech with voice cloning and a browser studio page
  • Qwen CPU engine runs on Windows, Linux x86-64 and macOS without Torch or CUDA

Weaknesses

  • Each engine has its own license and terms; users must verify them separately
  • Local playback needs PyAudio and PortAudio; Linux and macOS require a manual install
  • Qwen workflows require Python 3.11 or 3.12; CPU needs AVX2 on x86
  • Several engines need external setup such as mpv, a Piper binary or a separate server

What it needs

  • GPU optional
  • Docker
  • Needs PortAudio, mpv (some engines), NVIDIA driver (Qwen GPU mode)
  • Models: Qwen3-TTS, Inflect, Coqui XTTS, Kokoro, Piper
  • port 8080

What changed

From the release notes. No flag means the note does not say so, not that the upgrade is safe.

Also in Voice

See all 93
Also in Voice
RankProjectScore
1voicestudioDesktop app for voice cloning, dubbing, dictation and audiobooks113k starsAGPL-3.098 out of 100
2handyOffline push-to-talk dictation app that pastes transcripts into any text field39.1k starsMIT92 out of 100
3whisper.cppC/C++ speech-to-text engine running OpenAI Whisper models locally55.3k starsMIT90 out of 100
#4meetilyDesktop app that records, transcribes and summarizes meetings locally Live demo ↗ (opens in a new tab)32.8k starsMIT87 out of 100
#5buzzOffline desktop app for transcribing and translating audio with Whisper22.9k starsMIT84 out of 100