FireRedTeam

FireRedTeam / FireRedASR2S

Public

FireRedASR2S is a SOTA Industrial-Grade All-in-One ASR system with ASR, VAD, LID, and Punc modules. FireRedASR2 supports Chinese (Mandarin, 20+ dialects/accents), English, code-switching, and singing lyrics recognition. FireRedVAD supports speech/singing/music in 100+ langs. FireRedLID supports 100+ langs and 20+ zh dialects.

asr asr-pipeline audio-event-classification audio-event-detection automatic-speech-recognition
309
17
100% credibility
Found Feb 12, 2026 at 156 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

FireRedASR2S is a complete speech-to-text toolkit combining recognition, voice detection, language identification, and punctuation addition with top accuracy for Chinese dialects, English, and more.

How It Works

1
🔍 Discover FireRedASR2S

You hear about a super-smart tool that listens to audio and turns spoken words into accurate text, even handling tricky accents and dialects perfectly.

2
đź“Ą Get the tool ready

Download the simple package and set up a clean space for it on your computer with a few easy steps.

3
⬇️ Grab the smart listeners

Fetch the pre-trained helpers that understand speech, silence, languages, and add proper punctuation automatically.

4
🎵 Prepare your audio

Convert any sound clips to the right simple format if needed, so everything works smoothly.

5
🚀 Run the magic all-in-one

Feed your audio files into the full system, and watch it detect speech parts, identify languages, transcribe words, and add punctuation in one go.

6
đź“„ Review your results

Get back neatly formatted text with timings, confidence scores, and separated sentences ready to use.

âś… Perfect transcripts ready

Enjoy spot-on text from your audio, saving hours of manual work, whether for meetings, songs, or multilingual chats.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 156 to 309 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 FireRedASR2S?

FireRedASR2S is a Python-based, all-in-one ASR system that handles speech-to-text alongside voice activity detection (VAD), language identification (LID), and punctuation restoration (Punc). It processes 16kHz WAV files via CLI commands like `fireredasr2s-cli` or a simple Python API, outputting transcribed text with timestamps, confidence scores, language tags, and VAD segments in JSONL or SRT formats. Developers get an industrial-grade pipeline for turning raw audio into structured, punctuated transcripts without stitching separate tools.

Why is it gaining traction?

Its modules achieve SOTA performance—2.89% CER on Mandarin across four test sets and 97.57% F1 on multilingual VAD—outpacing FunASR, Whisper, and Doubao-ASR, especially on Chinese dialects and code-switching. Users notice low RTF (real-time factors under 0.1), word-level timestamps, and seamless integration for streaming or batch inference. Pretrained models download easily from Hugging Face or ModelScope, enabling quick prototyping without heavy setup.

Who should use this?

Speech engineers targeting Chinese apps—like dialect-heavy voice assistants, meeting transcription for Mandarin/English mixes, or lyrics recognition—will find it ideal. Backend devs building real-time ASR pipelines for call centers or podcasts get VAD/LID/Punc in one package, skipping fragmented open-source stacks. Avoid if you need non-Chinese focus or production-scale fine-tuning yet.

Verdict

Try FireRedASR2S if Chinese ASR is your bottleneck—benchmarks deliver real gains—but its 18 stars and 1.0% credibility score signal early-stage maturity with solid docs and examples. Solid for eval and prototypes; watch for community growth before deploying at scale.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.