ArchestackBest of Open-Source AI: List on GitHub ↗ (opens in a new tab)

#4444th of 51 in VoiceFor voice, speech and music

voicecraft

Zero-shot voice cloning TTS and speech editing from seconds of reference audio

Live demo ↗ (opens in a new tab)Website ↗ (opens in a new tab)Repository on GitHub ↗ (opens in a new tab)
Stars
8.6k
License
custom license
Last commit
May 2026
Language
Jupyter Notebook

Missed the latest GitHub check; these facts may be older.

Overview

VoiceCraft is a token-infilling neural codec language model that does zero-shot text-to-speech and speech editing, cloning an unseen voice from a few seconds of reference audio. The repo includes inference notebooks, command-line demos, a Gradio app on port 7860, and scripts for training and fine-tuning. Weights for 330M and 830M models are on Hugging Face.

Who it is for: Researchers and developers experimenting with voice cloning and speech editing

Strengths

  • Covers both zero-shot TTS and speech editing in one model
  • Gradio app with TTS, Edit and Long TTS modes, plus CLI scripts
  • Training and fine-tuning scripts and RealEdit manifests are included
  • Hosted demos on Hugging Face Spaces, Colab and Replicate

Weaknesses

  • Code is CC BY-NC-SA 4.0 and weights use Coqui license: non-commercial
  • Pinned old stack: Python 3.9, torch 2.0.1, CUDA 11.7, MFA, espeak
  • Fine-tuned models limit prompt plus generation to 16 seconds
  • Manual, heavy setup; the Docker route only launches Jupyter

What it needs

  • GPU required
  • Docker
  • Needs Python 3.9, PyTorch 2.0.1, audiocraft, ffmpeg, espeak-ng, Montreal Forced Aligner, Docker with NVIDIA container toolkit (optional)
  • Models: VoiceCraft giga330M, VoiceCraft giga830M, 330M/830M TTS Enhanced, Encodec 56M
  • port 7860

Also in Voice

See all 51
Also in Voice
RankProjectScore
1voicestudioDesktop app for voice cloning, dubbing and dictation with local speech models80.5k starsAGPL-3.083 out of 100
2xiaozhi-esp32ESP32 firmware for a voice chatbot with MCP device and cloud tool control31.7k starsMIT80 out of 100
3whisper.cppC/C++ port of OpenAI Whisper for local speech-to-text transcription54.5k starsMIT78 out of 100
#4VoiceboxLocal voice studio for cloning, TTS, dictation and agent speech58.2k starsMIT71 out of 100
#5buzzOffline desktop app for transcribing and translating audio with Whisper22.1k starsMIT71 out of 100