File size: 4,657 Bytes
a76ab92
ef4588e
 
 
 
 
 
 
 
 
 
 
 
a76ab92
ef4588e
 
 
 
 
ee96481
 
 
ef4588e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
64a91ba
ef4588e
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
---
license: mit
task_categories:
  - automatic-speech-recognition
language:
  - en
tags:
  - medical
  - asr
  - entity-cer
  - benchmark
size_categories:
  - n<1K
---

# MultiMed Hard — Medical ASR Benchmark

Entity-aware medical ASR benchmark — 50 hard rows from medical lectures and interviews.

Prepared by Trelis Research. Watch more on [Youtube](https://youtube.com/@TrelisResearch) or inquire about our custom voice AI (ASR/TTS) services [here](https://trelis.com/voice-ai-services).


## Source

Derived from [leduckhai/MultiMed](https://huggingface.co/datasets/leduckhai/MultiMed) EN test split (4,751 rows, MIT license). YouTube medical channels — lectures, interviews, podcasts, documentaries. Transcripts are human-reviewed.

## Preparation

1. Filter: audio ≥ 2s, ≤ 29s, text ≥ 20 chars
2. Casing filter: drop all-caps / all-lowercase rows
3. Whisper CER filter: drop rows with whisper-large-v3 CER > 10% (bad GT alignment)
4. Gemini Flash entity tagging (6 medical categories)
5. Keep rows with ≥ 1 entity (entity text ≥ 5 chars)
6. 3-model difficulty filter (whisper-large-v3, canary-1b-v2, Voxtral-Mini) with whisper-english normalization
7. Exclude median entity CER > 0.9
8. LLM validation (Gemini Flash): drop non-medical content, generic entities, GT typos
9. Top-50 by median entity CER

## Entity categories

- **drug** — drug or medication names (brand or INN)
- **condition** — diagnoses, diseases, syndromes, disorders
- **procedure** — surgical, diagnostic, or therapeutic procedures
- **anatomy** — anatomical structures, organs, body regions
- **biomarker** — lab tests, biomarkers, genes, proteins, molecular markers
- **organisation** — hospitals, regulatory bodies, pharmaceutical companies


## Columns

- `audio` — 16kHz WAV
- `text` — ground truth transcript (human-reviewed)
- `entities` — JSON array of tagged medical entities with `text`, `category`, `char_start`, `char_end`
- `difficulty_rank` — 1 = hardest
- `median_entity_cer` — median entity CER across 3 difficulty-filter models

## Leaderboard (16 models, sorted by Entity CER)

| # | Model | WER | CER | Entity CER | Results |
|---|---|---|---|---|---|
| 1 | scribe-v2 | 0.100 | 0.060 | 0.134 | [results](https://huggingface.co/datasets/Trelis/eval-scribe-v2-multimed-hard-20260408-1933) |
| 2 | MultiMed-ST (whisper-small-en) | 0.115 | 0.075 | 0.160 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-small-english-multimed-hard-20260408-1935) |
| 3 | gemini-2.5-pro | 0.105 | 0.062 | 0.167 | [results](https://huggingface.co/datasets/Trelis/eval-gemini-2.5-pro-multimed-hard-20260408-1933) |
| 4 | ursa-2-enhanced | 0.105 | 0.060 | 0.196 | [results](https://huggingface.co/datasets/Trelis/eval-ursa-2-enhanced-multimed-hard-20260408-1933) |
| 5 | whisper-large-v3 | 0.085 | 0.052 | 0.197 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-large-v3-multimed-hard-20260408-1932) |
| 6 | nova-3 | 0.120 | 0.069 | 0.199 | [results](https://huggingface.co/datasets/Trelis/eval-nova-3-multimed-hard-20260408-1934) |
| 7 | whisper-large-v3-turbo | 0.093 | 0.056 | 0.218 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-large-v3-turbo-multimed-hard-20260408-1931) |
| 8 | whisper-small | 0.133 | 0.075 | 0.228 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-small-multimed-hard-20260408-1933) |
| 9 | parakeet-tdt-0.6b-v3 | 0.159 | 0.101 | 0.233 | [results](https://huggingface.co/datasets/Trelis/eval-parakeet-tdt-0.6b-v3-multimed-hard-20260408-1930) |
| 10 | universal-3-pro | 0.125 | 0.100 | 0.234 | [results](https://huggingface.co/datasets/Trelis/eval-universal-3-pro-multimed-hard-20260408-1933) |
| 11 | canary-1b-v2 | 0.150 | 0.093 | 0.255 | [results](https://huggingface.co/datasets/Trelis/eval-canary-1b-v2-multimed-hard-20260408-1931) |
| 12 | whisper-v3 (fireworks) | 0.130 | 0.090 | 0.261 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-v3-multimed-hard-20260408-1936) |
| 13 | Voxtral-Mini-3B-2507 | 0.109 | 0.075 | 0.273 | [results](https://huggingface.co/datasets/Trelis/eval-Voxtral-Mini-3B-2507-multimed-hard-20260408-1931) |
| 14 | medasr | 0.251 | 0.145 | 0.278 | [results](https://huggingface.co/datasets/Trelis/eval-medasr-multimed-hard-20260409-1107) |
| 15 | whisper-tiny | 0.236 | 0.144 | 0.360 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-tiny-multimed-hard-20260408-1930) |
| 16 | whisper-base | 0.221 | 0.156 | 0.379 | [results](https://huggingface.co/datasets/Trelis/eval-whisper-base-multimed-hard-20260408-1930) |

Evaluated with [Trelis Studio](https://studio.trelis.com), whisper-english normalization.