vibevoice

A comprehensive ComfyUI integration for Microsoft's VibeVoice text-to-speech model, enabling high-quality single and multi-speaker voice synthesis directly within your ComfyUI workflows.

A ComfyUI custom node integration for local multi-engine multi-language Text-to-Speech and Voice Conversion. Supports: RVC, Echo-TTS, Qwen3-TTS, Cozy Voice 3, Step Audio EditX, IndexTTS-2, Chatterbox (classic and multilingual), F5-TTS, Higgs Audio 2, 3, and VibeVoice with unlimited text length, SRT timing, Character support, and many audio tools

ComfyUI custom node for the VibeVoice TTS. Expressive, long-form, multi-speaker conversational audio

VibeVoiceFusion is a full-stack, multi-speaker voice generation web system featuring LoRA fine-tuning, batch generation, and VRAM optimization. Based on Microsoft's VibeVoice (AR + diffusion architecture)

Audiobook creation app supporting too many TTS models (Qwen3-TTS, OmniVoice, VibeVoice, etc), focused on high-quality output. Plus audio-synced reader web app and standalone server component.

A multi-model local AI Voice Generator (TTS) studio. An open-source alternative to ElevenLabs and expensive online voice generators.

Rust implementation of VibeVoice text-to-speech with voice cloning and multi-speaker synthesis.

Beautiful voice app: record or upload to train a voice, generate speech from text or files, save & download voices.

Archive of the official Microsoft VibeVoice repository (7B & 1.5B). Backup of the deleted source code for the open-source TTS models, including the removed 7B version. Try the VibeVoice online service

ONNX speech pipeline library for ASR (diarization, VAD), and TTS

Real-time voice assistant on DGX Spark with 766ms latency to first audio

Create multi-voice podcasts with AI text-to-speech

Open-source models that turn Italian audio into text, locally, on your own hardware, with optional speaker diarization. Brought to you by the LocalAI team.

HOW TO RUN MICROSOFT VIBEVOICE LOCALLY

A Gradio-based demo for end-to-end vision-to-speech inference: Extract text or descriptions from images using Qwen2.5-VL-7B-Instruct, then convert to natural speech audio via Microsoft VibeVoice-Realtime-0.5B.

A ready-to-use Google Colab notebook for running the open-source VibeVoice TTS model from Microsoft, using the quantized Large Q8 variant (~12 GB VRAM) for multi-speaker long-form audio generation

WASA is an AI-powered transcription, translation and synthesis tool for voiceovers with independent channel control.

Production-ready Modal deployment of Microsoft VibeVoice-ASR (9B). Multi-hour audio via VAD-aware chunking, cross-chunk speaker unification with CAM++, auto-batched multi-GPU inference. 15x realtime.

Spanish voice stack (TTS + STT) for the homelab: VibeVoice-Realtime-0.5B at RTF 0.75 on CPU, plus Piper and whisper.cpp. Fully declared in NixOS, one-command deploy to Proxmox.