Open Source
Explore the latest AI open-source projects from GitHub and HuggingFace.
Explore the latest AI open-source projects from GitHub and HuggingFace.
FireRedTTS3 is a unified speech generation and editing system built on semantically enriched continuous speech representations, released by FireRedTeam. It ships in two variants: FireRedTTS3-Base for zero-shot voice cloning across 24 languages and 21 Chinese dialects, and FireRedTTS3-Instruct for natural-language voice design and speech editing. The Instruct variant can synthesize an entirely new voice from a text description alone with no reference audio required, and supports semantic edits (word insertion, deletion, substitution) and acoustic edits (speed, pitch, volume) through a single instruction-driven API. On the Seed-TTS-eval benchmark, FireRedTTS3-Base reports the best average WER/SIM scores among the models it compares against, including Qwen3-TTS, VoxCPM2, and dots.tts. Pretrained weights are published on both Hugging Face and ModelScope alongside a technical report.
RVC-Boss
Open-source WebUI for few-shot and zero-shot voice cloning and text-to-speech, producing a usable voice from as little as a 5-second sample.
Microsoft
Microsoft's MIT-licensed open frontier voice AI: 1.5B long-form TTS up to 90 minutes with 4 speakers, 0.5B streaming TTS at 300 ms latency, and 7B ASR for 60-minute single-pass transcription. 47k+ stars.
2noise
A dialogue-optimized open TTS model trained on 100,000+ hours that adds fine-grained prosody — laughter, pauses, interjections — with multi-speaker, English/Chinese support.
suno-ai
Suno's fully generative text-to-audio model — speech, music, and sound effects from one transformer, with nonverbal cues like [laughs] and 100+ voice presets (MIT).