Repository navigation
Conversation
… boosting) Qwen3-ASR runs through whisperkitpro-cli and is selected by --model name alone: the CLI detects it by prefix and resolves the model repo itself, and a --model-prefix glob would stop that detection from matching. The engine gains an is_qwen3_asr mode (model_version only) alongside model-path and legacy modes, and a whisperkitpro-qwen3-asr alias is registered. Keyword boosting (--use-keywords) flows through the existing custom-vocabulary path and is applied via Qwen3-ASR's native system-prompt biasing. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
With --model alone whisperkitpro-cli resolves the package from the Hub on every call (one call per sample), and parallel runs were rate-limited (HTTP 429). The alias now downloads argmaxinc/qwenasrkit-pro qwen3-asr/ once and passes it as --model-path; --model qwen3-asr-1.7b still selects the Qwen3-ASR arm.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
A
whisperkitpro-qwen3-asr-1.7btranscription pipeline: Qwen3-ASR 1.7B throughwhisperkitpro-cli.src/openbench/engine/whisperkitpro_engine.py): amodel_versionstarting withqwen3-asris always passed as--model <name>:whisperkitpro-clidetects Qwen3-ASR from that name (a--model-pathfolder alone would load as Whisper, and a--model-prefixglob would stop the detection). Withrepo_id+model_variant(ormodel_dir) the package is also passed as--model-path; withmodel_versionalone the CLI downloads it itself, once per call.whisperkitpro-qwen3-asr-1.7b: downloadsargmaxinc/qwenasrkit-proqwen3-asr/once and passes--model qwen3-asr-1.7b --model-path <dir>(per-call Hub resolution was rate-limited, HTTP 429, across parallel runs).--use-keywordsworks: Qwen3-ASR applies the keywords through its system prompt.model_dirprecedence (tests/engine/test_whisperkitpro_engine.py).Qwen3-ASR needs macOS 26 or later on Apple silicon;
whisperkitpro-clirefuses older macOS.Validation
tests/engine/test_whisperkitpro_engine.py: 10 passed.openbench-cli evaluate -p whisperkitpro-qwen3-asr-1.7bon macOS 26 (first samples): librispeech-200 WER 0.85% (10), common-voice-es WER 3.53% (10), earnings22-kws-chunkwise WER 10.8% / keyword F1 0.71 with--use-keywords, WER 13.1% / F1 0.36 without (5).