Official documentation
FireRedASR2S: README
FireRedTeam, GitHub, FireRedTeam/FireRedASR2S, 2026
Read the originalRead on October 20, 2026
- What it is
- The README of FireRedTeam's FireRedASR2S repository, which releases the FireRedASR2 speech recognition models with their model weights and inference code. Its Available Models and Languages table lists the languages and dialects each model supports.
- How much weight it carries
- The maker's own repository, read on 2026-10-20. The table is the project's stated support list for the recognition models. It names dialects and code-switching, which are not counted as languages here, and it gives 100+ languages only for the language-identification and voice-activity models, which do not transcribe speech. The README says a technical report is still to come, so the figures may change.
Facts taken from this source
- FireRedTeam's FireRedASR2 recognition models, FireRedASR2-LLM and FireRedASR2-AED, support two languages, Chinese (Mandarin, with 20+ dialects and accents) and English, plus code-switching.
“Model Supported Languages & Dialects Download FireRedASR2-LLM Chinese (Mandarin and 20+ dialects/accents * ), English, Code-Switching” (Available Models and Languages table)