FireRedASR is an open-source industrial-grade automatic speech recognition (ASR) system supporting Mandarin, Chinese dialects, and English. It provides two model variants: FireRedASR-AED (Attention-based Encoder-Decoder with Conformer encoder and Transformer decoder) and FireRedASR-LLM (Encoder-Adapter-LLM framework leveraging Qwen2-7B). The codebase includes a CLI tool (`speech2text.py`), a Python API (`FireRedAsr`), feature extraction (fbank with CMVN), tokenization (mixed Chinese/English SentencePiece), and utilities for WER computation. It also includes a complete Triton Inference Server deployment with TensorRT-LLM acceleration for the AED model, with Docker support and benchmarking scripts. The architecture is modular: data preprocessing, model definitions, tokenizers, and inference orchestration are cleanly separated, though the model components are tightly coupled to the specific architectures. The repository is actively maintained with recent commits and demonstrates strong engineering craftsmanship in both the core ASR implementation and the production-grade deployment stack.
This repository is a curated collection of Chinese vocabulary lists for standardized tests (HSK 3.0, BCT, YCT), enriched with pinyin, zhuyin, part-of-speech tags, frequency data, and English definitions. It includes Python and JavaScript scripts for data cleaning, format conversion (TXT/TSV/XLSX), POS tag correction using SUBTLEX frequency corpus, and pinyin tone conversion. The data is organized by test level and version (2021 and 2025 HSK standards), with multiple output formats suitable for flashcard apps (Anki, Pleco) and language learning tools. The repository is primarily a data asset with supporting ETL scripts, not a production application. It demonstrates solid data engineering practices for language resource curation, though it suffers from some file duplication and hardcoded paths.