Pith. sign in

REVIEW 8 cited by

FireRedASR: Open-Source Industrial-Grade Mandarin Speech Recognition Models from Encoder-Decoder to LLM Integration

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2501.14350 v1 pith:SRILMXVW submitted 2025-01-24 eess.AS cs.SD

Kai-Tuo Xu , Feng-Long Xie , Xu Tang , Yao Hu This is my paper · ORCID
classification eess.AScs.SD
keywords speechmandarinmodelsfireredasrbenchmarksdesignedfireredasr-llmparameters
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

We present FireRedASR, a family of large-scale automatic speech recognition (ASR) models for Mandarin, designed to meet diverse requirements in superior performance and optimal efficiency across various applications. FireRedASR 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. On public Mandarin benchmarks, FireRedASR-LLM (8.3B parameters) achieves an average Character Error Rate (CER) of 3.05%, surpassing the latest SOTA of 3.33% with an 8.4% relative CER reduction (CERR). It demonstrates superior generalization capability over industrial-grade baselines, achieving 24%-40% CERR in multi-source Mandarin ASR scenarios such as video, live, and intelligent assistant. 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. On public Mandarin benchmarks, FireRedASR-AED (1.1B parameters) achieves an average CER of 3.18%, slightly worse than FireRedASR-LLM but still outperforming the latest SOTA model with over 12B parameters. It offers a more compact size, making it suitable for resource-constrained applications. Moreover, both models exhibit competitive results on Chinese dialects and English speech benchmarks and excel in singing lyrics recognition. To advance research in speech processing, we release our models and inference code at https://github.com/FireRedTeam/FireRedASR.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. LLMs and Speech: Integration vs. Combination

    eess.AS 2026-03 unverdicted novelty 6.0 of 10

    With matched data and sizes, CTC+LLM shallow fusion beats tight speech-LLM integration on in-domain ASR, while prefix LLMs win average WER on out-of-domain HuggingFace sets.

  2. SegTune: Structured and Fine-Grained Control for Song Generation

    cs.SD 2025-10 conditional novelty 6.0 of 10

    SegTune generates songs where each musical section follows its own text description, using a fine-tuned LLM to predict lyric timings so per-section instructions land in the correct audio window.

  3. WenetSpeech-Yue: A Large-scale Cantonese Speech Corpus with Multi-dimensional Annotation

    cs.SD 2025-09 conditional novelty 6.0 of 10

    The authors built and released the largest open-source Cantonese speech corpus (21,800 hours, 10 domains, rich metadata), and show that models trained on it match or beat existing speech recognition and synthesis systems.

  4. Existence of the longest arcs for left-invariant three-dimensional contact sub-Lorentzian structures

    math.DG 2026-03 unverdicted novelty 5.0 of 10

    Sufficient conditions are given for existence of longest arcs in left-invariant three-dimensional contact sub-Lorentzian structures on solvable Lie groups and the universal cover of SL(2,R).

  5. FireRedChat: A Pluggable, Full-Duplex Voice Interaction System with Cascaded and Semi-Cascaded Implementations

    cs.SD 2025-09 conditional novelty 5.0 of 10

    A full-duplex voice system with streaming personalized VAD and semantic end-of-turn detection reports fewer false barge-ins and latencies near commercial benchmarks.

  6. Cross-Learning Fine-Tuning Strategy for Dysarthric Speech Recognition Via CDSD database

    cs.SD 2025-08 conditional novelty 5.0 of 10

    Joint fine-tuning on seven dysarthric speakers' data reduced per-speaker character error rates by up to 13.15 percentage points compared to single-speaker fine-tuning on the CDSD corpus.

  7. Mixture of LoRA Experts with Multi-Modal and Multi-Granularity LLM Generative Error Correction for Accented Speech Recognition

    cs.SD 2025-07 conditional novelty 5.0 of 10

    A three-stage pipeline of mono-accent LoRA experts, hierarchical routing, and phoneme-plus-word LLM error correction cuts accented-English WER from 6.34% to 2.07% on a combined 9-accent test set.

  8. Breaking the Barriers of Text-Hungry and Audio-Deficient AI

    cs.SD 2025-06 reject novelty 4.0 of 10

    A proposed audio-native translation framework called MAST with fractional diffusion is described, but no evidence is given that it produces working translations.

Pith tools