FireRedASR2 AED community proposal: local transcription endpoint and chunking scope
Proposing a community integration note for the FireRedASR2 AED MLX pack through Rapid-MLX's local transcription endpoint.
The catalog identifies the AED family, and the STT adapter contains a FireRed-specific long-audio chunking path. The proposed example should follow the card's 16 kHz mono input guidance and verify transcript ordering across chunks, not just load the model.
It should distinguish this batch transcription path from the broader FireRedASR2S streaming system. A future ready-to-run recipe needs a pinned audio sample, transcript result, runtime versions and measured resource use; no streaming-latency or recognition-quality claim is justified by catalog presence. The source and catalog contracts below identify the concrete integration to review.
Concrete catalog entry: fireredasr2-aed → mlx-community/FireRedASR2-AED-mlx@f3212eacfa49b851130b97c63653c8e06ee09bdb. This is a conversion/distribution artifact; it is not presented as an official upstream release.
Source references (fixed Rapid commit): catalog, STT/chunking adapter, catalog contracts, audio API guide.
Verification status: this proposal is based on reading the catalog, implementation and the linked documentation/tests. No fresh model inference or test-suite run was performed for this submission; source tests are not an end-to-end qualification. It requests a community documentation scope, not an official endorsement or a performance listing.
Disclosure: I maintain Rapid-MLX. Prepared with AI assistance under my direction. Anonymous usage-metadata telemetry is enabled by default; DO_NOT_TRACK=1 disables it. The original model and conversion terms continue to apply.