Knowledge Share
Technical articles, tutorials, and insights
Open Source Project of the Day (#139): Voicebox — A Local, Open-Source Alternative to ElevenLabs
Voicebox is a local-first open-source AI voice studio with 7 TTS engines, zero-shot voice cloning, a global dictation hotkey, and MCP agent voice integration. Supports 23 languages. Voice data never leaves your device. 32.3k Stars, MIT license.
One Open Source Project a Day (No.86): VibeVoice - Microsoft's Speech AI That Processes 90 Minutes of Audio in a Single Pass
A deep dive into VibeVoice, Microsoft Research's open-source speech AI family with 44.9k+ Stars and an ICLR 2026 Oral award. A 7.5 Hz ultra-low framerate tokenizer compresses 90 minutes of audio to just 40,500 tokens, enabling end-to-end LLM reasoning over full-length audio. Three models cover multi-speaker TTS (90 min), long-audio ASR (60 min), and real-time streaming synthesis (300ms first-chunk latency). MIT licensed — with a cautionary note on misuse.
One Open Source Project a Day (Part 60): IndexTTS - Bilibili's Industrial Zero-Shot Text-to-Speech System
Deep dive into IndexTTS, Bilibili IndexTeam's open-source zero-shot TTS system — emotion control, voice cloning, duration control; IndexTTS2 decouples emotion and timbre, multi-modal emotion input, industrial-grade controllability
Open Source Project of the Day (Part 12): SoulX-Podcast - Multi-Turn Conversational Podcast Generation for More Natural and Authentic AI Speech
A deep dive into SoulX-Podcast, an open-source podcast generation system with 3.1k+ Stars — supporting multi-turn multi-speaker dialogue, cross-dialect zero-shot voice cloning, and paralinguistic control for more natural and authentic AI-generated speech
Open Source Project of the Day (Part 11): Supertonic - Lightning-Fast On-Device Multilingual TTS, Running Speech Synthesis Locally at Full Speed
A deep dive into Supertonic, an open-source on-device TTS system with 2.6k+ Stars — achieving blazing-fast speech synthesis via ONNX Runtime, supporting 5 languages, with built-in text normalization, no cloud required, running entirely on-device