Open Source

Explore the latest AI open-source projects from GitHub and HuggingFace.

Evermx

Latest AI/LLM news and in-depth reviews.
We analyze usability, potential, and trade-offs.

info@evermx.com

LLM

Claude
Gemini
GPT
Llama
Other LLM

Official Sites

Anthropic (Claude)
Google AI (Gemini)
OpenAI (GPT)
Meta AI (Llama)
Hugging Face

Reviews Tools Open Source Live Profile

VibeVoice - Open Source | Evermx | Evermx

Back to Open Source

Trending

VibeVoice

microsoftMIT

View on GitHub

TTS23.5K Stars2.6K Forks10 views

Microsoft's open-source frontier voice AI combining text-to-speech and automatic speech recognition. Features 60-minute long-form ASR, 90-minute continuous TTS, and real-time streaming with 300ms latency using continuous speech tokenizers at 7.5 Hz frame rate.

Key Features

60-minute long-form ASR in a single pass with speaker identification
90-minute continuous TTS with up to 4 distinct speakers
Real-time streaming TTS with 0.5B parameter model and 300ms latency
Multilingual support covering 50+ languages
Ultra-low 7.5 Hz continuous speech tokenizer with next-token diffusion

Related Projects

TrendingTTS

GitHub

14.5K1.5K

Sesame CSM

Sesame AI Labs

Sesame AI Labs' open-source 1B-parameter conversational speech model using Llama architecture — natural human-like intonation, multi-speaker support, HuggingFace Transformers native.

Kokoro TTS

hexgrad

Lightweight 82M parameter open-weight TTS model delivering quality comparable to larger models, with Apache 2.0 licensing and 9-language support.

Apache-2.013