FireRedTeam 开源工业级 ASR 模型 FireRedASR，普通话基准达新 SOTA
GitHub - FireRedTeam/FireRedASR: Open-source industrial-grade ASR models supporting Mandarin, Chinese dialects and English, achieving a new SOTA on public Mandarin ASR benchmarks, while also offering outstanding singing lyrics recognition capability. · GitHub
FireRedTeam 开源工业级 ASR 模型 FireRedASR，支持普通话、中文方言和英语，在公开普通话 ASR 基准上取得新 SOTA，并具备出色的歌词识别能力。该系列包含基于 Encoder-Adapter-LLM 框架的 FireRedASR-LLM 和基于 AED 架构的 FireRedASR-AED 两个版本。FireRedASR-AED 支持最长 60s 音频输入，FireRedASR-LLM 支持最长 30s。
正文
当前显示原文，中文全文译文尚未就绪。
FireRedTeam; FireRedASR
Repository navigation
NotificationsYou must be signed in to change notification settings; Fork167 (167); Star2k (2k)You must be signed in to star a repository
Open-source industrial-grade ASR models supporting Mandarin, Chinese dialects and English, achieving a new SOTA on public Mandarin ASR benchmarks, while also offering outstanding singing lyrics recognition capability.
Repository files navigation
README; Apache-2.0 license
[Paper] [Model] [Blog] [Demo]
FireRedASR2S has been open-sourced! Welcome to try it! https://github.com/FireRedTeam/FireRedASR2S
FireRedASR2S is a state-of-the-art (SOTA), industrial-grade, all-in-one ASR system with ASR, VAD, LID, and Punc modules. All modules achieve SOTA performance.
FireRedASR is a family of open-source industrial-grade automatic speech recognition (ASR) models supporting Mandarin, Chinese dialects and English, achieving a new state-of-the-art (SOTA) on public Mandarin ASR benchmarks, while also offering outstanding singing lyrics recognition capability.
[2026.02.25] We release FireRedASR2-LLM model weights. 🤗 🤖; [2026.02.12] We release FireRedASR2S (FireRedASR2-AED, FireRedVAD, FireRedLID, and FireRedPunc) with model weights and inference code. See https://github.com/FireRedTeam/FireRedASR2S; [2025.02.17] We release FireRedASR-LLM-L model weights; [2025.01.24] We release technical report, blog, and FireRedASR-AED-L model weights
Method
FireRedASR is designed to meet diverse requirements in superior performance and optimal efficiency across various applications. It comprises two variants:
FireRedASR-LLM: Designed to achieve state-of-the-art (SOTA) performance and to enable seamless end-to-end speech interaction. It adopts an Encoder-Adapter-LLM framework leveraging large language model (LLM) capabilities; FireRedASR-AED: Designed to balance high performance and computational efficiency and to serve as an effective speech representation module in LLM-based speech models. It utilizes an Attention-based Encoder-Decoder (AED) architecture
Evaluation
Results are reported in Character Error Rate (CER%) for Chinese and Word Error Rate (WER%) for English.
ws means WenetSpeech.
Usage
Download model files from huggingface and place them in the folder pretrained_models.
If you want to use FireRedASR-LLM-L, you also need to download Qwen2-7B-Instruct and place it in the folder pretrained_models. Then, go to folder FireRedASR-LLM-L and run $ ln -s ../Qwen2-7B-Instruct
Setup
Create a Python environment and install dependencies
Set up Linux PATH and PYTHONPATH
Convert audio to 16kHz 16-bit PCM format
When performing batch beam search with FireRedASR-LLM, please ensure that the input lengths of the utterances are similar. If there are significant differences in utterance lengths, shorter utterances may experience repetition issues. You can either sort your dataset by length or set batch_size to 1 to avoid the repetition issue
Input Length Limitations
FireRedASR-AED supports audio input up to 60s. Input longer than 60s may cause hallucination issues, and input exceeding 200s will trigger positional encoding errors; FireRedASR-LLM supports audio input up to 30s. The behavior for longer input is currently unknown
Acknowledgements
Thanks to the following open-source works:
Citation
Open-source industrial-grade ASR models supporting Mandarin, Chinese dialects and English, achieving a new SOTA on public Mandarin ASR benchmarks, while also offering outstanding singing lyrics recognition capability.