Skip to content

Add Qwen3-ASR 1.7B Keyword Boosting Results #102

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
atiorh merged 2 commits into from
Aug 5, 2026
Merged

Add Qwen3-ASR 1.7B Keyword Boosting Results #102

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
51e90c9
BENCHMARKS: add Qwen3-ASR 1.7B keyword recognition results
Aug 4, 2026
82d55a8
BENCHMARKS: fill Qwen3-ASR no-keywords baseline column
Aug 5, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 11 additions & 0 deletions BENCHMARKS.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -505,6 +505,13 @@
- **Code Reference:** [openbench/pipeline/transcription/transcription_whisperkitpro.py](https://github.com/argmaxinc/OpenBench/blob/main/src/openbench/pipeline/transcription/transcription_whisperkitpro.py)
- **Hardware**: M2 Ultra Mac Studio

### Argmax
- **Latest Run:** `2026-08-04`
- **Model Version:** `qwen3-asr-1.7b`
- **Configuration:** Argmax SDK WhisperKit Pro framework with the Qwen3-ASR 1.7B model and Custom Vocabulary feature enabled. Keywords are applied through Qwen3-ASR's native system-prompt context biasing.
- **Code Reference:** [openbench/pipeline/transcription/transcription_whisperkitpro.py](https://github.com/argmaxinc/OpenBench/blob/main/src/openbench/pipeline/transcription/transcription_whisperkitpro.py)
- **Hardware**: M3 Pro MacBook Pro

### Apple
- **Latest Run:** `2025-12-30`
- **Model Version:** `SFSpeechRecognizer`
Expand Down Expand Up @@ -557,6 +564,7 @@
| Whisper OSS<br/>(large-v3-turbo) | 15.4 | 21.24 | 14.69 |
| Argmax<br/>(parakeet-v2) | 14.69 | 12.46 | 12.57 |
| Argmax<br/>(parakeet-v3) | 16.89 | 14.57 | 14.73 |
| Argmax<br/>(qwen3-asr-1.7b) | 11.86 | 9.19 | 10.18 |
| ElevenLabs | 10.53 | 9.13 | 9.08 |
| Apple<br/>(SFSpeechRecognizer) | 28.42 | 26.98 | 27.26 |
| Apple<br/>(SpeechAnalyzer) | 17 | - | - |
Expand Down Expand Up @@ -590,6 +598,7 @@ If the model predicts 20 keywords and 15 of them match the ground truth, precisi
| Whisper OSS<br/>(large-v3-turbo) | 0.97 | 0.96 | 0.94 |
| Argmax<br/>(parakeet-v2) | 0.97 | 0.98 | 0.96 |
| Argmax<br/>(parakeet-v3) | 0.98 | 0.98 | 0.95 |
| Argmax<br/>(qwen3-asr-1.7b) | 0.97 | 0.98 | 0.93 |
| ElevenLabs | 0.97 | 0.99 | 0.96 |
| Apple<br/>(SFSpeechRecognizer) | 1 | 0.99 | 0.99 |
| Apple<br/>(SpeechAnalyzer) | 0.99 | - | - |
Expand Down Expand Up @@ -623,6 +632,7 @@ If the ground-truth transcript has 25 keywords and the model correctly finds 15,
| Whisper OSS<br/>(large-v3-turbo) | 0.53 | 0.77 | 0.82 |
| Argmax<br/>(parakeet-v2) | 0.47 | 0.85 | 0.82 |
| Argmax<br/>(parakeet-v3) | 0.45 | 0.82 | 0.8 |
| Argmax<br/>(qwen3-asr-1.7b) | 0.53 | 0.87 | 0.81 |
| ElevenLabs | 0.75 | 0.96 | 0.94 |
| Apple<br/>(SFSpeechRecognizer) | 0.26 | 0.45 | 0.4 |
| Apple<br/>(SpeechAnalyzer) | 0.39 | - | - |
Expand Down Expand Up @@ -658,6 +668,7 @@ F1 = 2 × (0.75 × 0.6) / (0.75 + 0.6) = **66.7%**, reflecting the model's overa
| Whisper OSS<br/>(large-v3-turbo) | 0.69 | 0.86 | 0.87 |
| Argmax<br/>(parakeet-v2) | 0.63 | 0.91 | 0.88 |
| Argmax<br/>(parakeet-v3) | 0.62 | 0.89 | 0.87 |
| Argmax<br/>(qwen3-asr-1.7b) | 0.69 | 0.92 | 0.87 |
| ElevenLabs | 0.84 | 0.97 | 0.95 |
| Apple<br/>(SFSpeechRecognizer) | 0.41 | 0.62 | 0.58 |
| Apple<br/>(SpeechAnalyzer) | 0.56 | - | - |
Expand Down
Loading