mudler

mudler / vibevoice.cpp

Public

C++ port of Microsoft VibeVoice built on ggml

13
0
100% credibility
Found Apr 29, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
C++
AI Summary

A local C++ tool for generating realistic speech from text using voice cloning and transcribing long audio into speaker-labeled transcripts.

How It Works

1
🔍 Discover Local Voice Magic

You stumble upon this handy tool that lets you turn everyday text into realistic speech or transcribe conversations perfectly, all running on your own computer without needing fancy setups.

2
📥 Grab the Essentials

Download the simple program and a few ready-to-use voice samples from the shared collection to get started right away.

3
🎤 Speak Your Words

Type in a sentence, pick a voice, and with one command, hear it come alive in that exact voice, feeling like magic as your text turns into natural-sounding audio.

4
Choose Your Adventure
📢
Make More Speech

Keep crafting custom voice messages or stories effortlessly.

🎧
Transcribe Audio

Upload recordings and get timed transcripts with who spoke when.

5
⚡ Test Round Trip

Try converting speech back to text to see how spot-on and reliable it is.

✅ Voice Dreams Realized

Now you can clone voices, generate audio anytime, or transcribe meetings perfectly, all locally and super fast on your machine.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 13 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is vibevoice.cpp?

vibevoice.cpp is a C++ port of Microsoft's VibeVoice, delivering local TTS with voice cloning and ASR for long-form transcription plus speaker diarization. Built on ggml, it runs quantized models from Hugging Face in a single binary—no Python needed—via simple CLI commands like `./vibevoice-cli tts --text "Hello" --out hello.wav` or ASR on WAV files. Developers get CPU/GPU inference (CUDA, Metal) with C++ github actions for builds and a flat C API for embedding in C++ github projects or Go apps.

Why is it gaining traction?

It stands out with ggml's free backend support, hitting 0.4 RTF on GPU for ASR benchmarks, and pre-quantized models that slash disk use by 50% without quality loss. The closed-loop TTS→ASR test validates end-to-end fidelity, and scripts let you convert upstream models or voices yourself. For c++ github copilot fans or c++ portable IDE users, it's a drop-in for vibevoice cpp without Transformers overhead.

Who should use this?

AI engineers integrating voice into desktop apps or edge devices, LocalAI users extending to VibeVoice, or c++ github projects for beginners tackling TTS/ASR prototypes. Perfect for c++ portfolio projects needing real-time cloning or diarization, like podcast tools or voice assistants.

Verdict

Early alpha (13 stars, 1.0% credibility) with solid tests and docs, but needs more voices/models—PRs welcome. Grab it for c++ github trending experiments if you're in local AI; skip for production until mature.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.