Whisper medium MN — Mongolian Speech-to-Text Benchmark
Real results from the Whisper medium MN Batch API (enhanced operating point, language mn) evaluated across 4 Mongolian datasets and 265 audio samples. Every number below is measured — not marketing. WER, speed, and pricing are shown as-is, brutally honest.
265/265 samples transcribed · 100% success rate
Lower is better · across 265 samples
Pricing
Whisper medium MN list pricing for batch transcription. No discounts, no negotiated rates applied — the raw per-minute rate.
Pricing source: Whisper medium MN public pricing. Duudlaga Flow is shown for context only — this page isolates Whisper medium MN so the number is not padded by our own product.
WER & CER by dataset
Word and character error rates per dataset. Lower is better — and these are the real Whisper medium MN numbers, which are weak on Common Voice 24.
Dataset summary
Aggregate accuracy, speed, and timing for each dataset.
| Dataset | Source | Samples | Success | WER | CER | Accuracy | Speed | Audio (s) | Proc (s) |
|---|---|---|---|---|---|---|---|---|---|
| Common Voice 24 (MN) | Hugging Face → | 59 | 59/59 | 9.9% | 3.2% | 90.1% | 2.64× | 330.6s | 125.1s |
| Shunya Labs Mongolian Speech | Hugging Face → | 60 | 60/60 | 2.8% | 1.5% | 97.2% | 2.94× | 645.1s | 219.6s |
| Common Voice 20 (MN) | Hugging Face → | 54 | 54/54 | 32.3% | 10.5% | 67.7% | 2.33× | 269.8s | 115.7s |
| Modern Voice | — | 92 | 92/92 | 37.7% | 16.3% | 62.3% | 2.68× | 603.8s | 225.4s |
WER vs speed — per sample
Each dot is one audio sample. The sweet spot is the bottom-left (low error, fast). Whisper medium MN sits high on error for many Common Voice 24 samples.
Per-sample results
Ground truth shown verbatim in the Expected column. The result column highlights only the words Whisper medium MN got wrong, in red — no strikethrough/swap gymnastics, just the mistakes.
| # | Audio | Sample | Dataset | Expected (ground truth) | Whisper medium MN result | WER | CER | I/D/S |
|---|---|---|---|---|---|---|---|---|
| 1 | btsee_0001 | Common Voice 24 (MN) | Гэхдээ амьсгал хураахаасаа өмнө танд мэдэж байгаагаа хэлье. | гэхдээ амьсгал хураахаасаа өмнө танд мэдэж байгаагаа хэлье | 0.0% | 0.0% | 0/0/0 | |
| 2 | btsee_0002 | Common Voice 24 (MN) | Надад заяасан аз жаргал гэдэг ердөө гуравхан сарын хугацаатай байсан гэж үү? | надад заяасан аз жаргал гэдэг ердөө гуравхан сарын хугацаатай байсан гэж үү | 0.0% | 0.0% | 0/0/0 | |
| 3 | btsee_0003 | Common Voice 24 (MN) | Одоо бид өөрсдөө өвчин эмгэгээсээ салахыг хичээцгээе. | одоо бид өөрсдөө өвчин эмгээгүйсээ салахыг хичээцгээ | 28.6% | 7.7% | 0/0/2 | |
| 4 | btsee_0004 | Common Voice 24 (MN) | Би бол голдуу хээрээр гэр, хэцээр дэр хийж явдаг хүн. | би бол голдуу хээрээр гэр хэцээр дэр хийж явдаг хүн | 0.0% | 0.0% | 0/0/0 | |
| 5 | btsee_0005 | Common Voice 24 (MN) | Хан хурмаст уурлаж, Болдоггүй Бор өвгөнийг хор луугаараа ниргүүлэхээр явуулжээ. | хан хурмаст уурлаж болдоггүй бор өвгөнийг хоёр луугаараа ниргүүлэхээр явуулжээ | 10.0% | 1.3% | 0/0/1 | |
| 6 | btsee_0006 | Common Voice 24 (MN) | Алив наашаа ороод ир гээд гэртээ оров. | алив наашаа ороод ир гээд гэртээ оров | 0.0% | 0.0% | 0/0/0 | |
| 7 | btsee_0007 | Common Voice 24 (MN) | Өө өндөр дээдэс таны тухайд би баталж чадахгүй. | өө өндөр дээдэс таны тухайд би баталж чадахгүй | 0.0% | 0.0% | 0/0/0 | |
| 8 | btsee_0008 | Common Voice 24 (MN) | Харин гурав дахь удаагаас эхлэн хүмүүсийг сонирхож эхлэв. | харин гурав дахь удаагаас эхлэх хүмүүсийг сонирхож эхлэв | 12.5% | 1.8% | 0/0/1 | |
| 9 | btsee_0009 | Common Voice 24 (MN) | Та нар очингуутаа шөл л өгч үз. | та нар очингуутаа шөлөл өгч үз | 28.6% | 3.3% | 0/1/1 | |
| 10 | btsee_0010 | Common Voice 24 (MN) | Ерөөсөө литр үйлдвэрээсээ салаагүй явсан юм чинь. | ирвэс уйлдвэрлэсээ салаагүй явсан юм чинь | 42.9% | 25.0% | 0/1/2 |
Methodology
How these numbers were produced.
Provider: Whisper medium MN (Whisper medium fine-tuned for Mongolian on a private corpus — the second most-downloaded Mongolian Whisper fine-tune.). Models with this badge were trained on data that overlaps the benchmark corpora (Mongolian Common Voice — 113 of 173 samples — or the public Shunya Labs corpus). Scores on overlapping corpora are inflated by memorization; judge these models by the per-dataset table on their details page, especially the corpora they were NOT trained on.
Endpoint local (transformers). Language mn. Operating point Cafet/whisper-meduim-mongolian (Mongolian fine-tune) on mps. Diarization none.
Datasets: Common Voice 24 (MN), Shunya Labs Mongolian Speech, Common Voice 20 (MN), Modern Voice — 265 samples, 1849.3s of audio total.
- Common Voice 24 (MN)https://huggingface.co/datasets/btsee/common-voices-24-mn
- Shunya Labs Mongolian Speechhttps://huggingface.co/datasets/shunyalabs/mongolian-speech-dataset
- Common Voice 20 (MN)https://huggingface.co/datasets/warmestman/common-voice-20-mn-normalized
- Modern Voice—
Metrics: WER and CER are computed with a standard word/character Levenshtein alignment, normalized for case and punctuation. Accuracy = 100 − WER. Speed factor = audio duration ÷ processing time (× realtime). All requests are real Whisper medium MN Batch API calls, not cached or simulated.
Diff highlighting: The result column aligns to the ground truth and colors every substitution and insertion red. Deletions (words missing from the result) are not shown in the result column — the Expected column already holds the full ground truth as-is.