Whisper large-v3 MN — Mongolian Speech-to-Text Benchmark
Real results from the Whisper large-v3 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 large-v3 MN list pricing for batch transcription. No discounts, no negotiated rates applied — the raw per-minute rate.
Pricing source: Whisper large-v3 MN public pricing. Duudlaga Flow is shown for context only — this page isolates Whisper large-v3 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 large-v3 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 | 40.1% | 13.1% | 59.9% | 1.51× | 330.6s | 218.4s |
| Shunya Labs Mongolian Speech | Hugging Face → | 60 | 60/60 | 28.6% | 8.8% | 71.4% | 1.76× | 645.1s | 367.1s |
| Common Voice 20 (MN) | Hugging Face → | 54 | 54/54 | 41.3% | 15.8% | 58.7% | 1.34× | 269.8s | 200.9s |
| Modern Voice | — | 92 | 92/92 | 44.9% | 19.3% | 55.1% | 1.56× | 603.8s | 388.2s |
WER vs speed — per sample
Each dot is one audio sample. The sweet spot is the bottom-left (low error, fast). Whisper large-v3 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 large-v3 MN got wrong, in red — no strikethrough/swap gymnastics, just the mistakes.
| # | Audio | Sample | Dataset | Expected (ground truth) | Whisper large-v3 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 хэлээ | 12.5% | 3.4% | 0/0/1 | |
| 2 | btsee_0002 | Common Voice 24 (MN) | Надад заяасан аз жаргал гэдэг ердөө гуравхан сарын хугацаатай байсан гэж үү? | надад заяасан аж жаргал гэдэг эрдээ гуравхан сарын хугацаатай байсан гэж үү | 16.7% | 5.3% | 0/0/2 | |
| 3 | btsee_0003 | Common Voice 24 (MN) | Одоо бид өөрсдөө өвчин эмгэгээсээ салахыг хичээцгээе. | одоо би дөөрч дөө өвчин эмгээгээсээ салахыг чивцгээв | 71.4% | 19.2% | 1/0/4 | |
| 4 | btsee_0004 | Common Voice 24 (MN) | Би бол голдуу хээрээр гэр, хэцээр дэр хийж явдаг хүн. | би бол голдуу хээрээр гэр хэцээр дэрхий ч явдаг хүн | 20.0% | 5.9% | 0/0/2 | |
| 5 | btsee_0005 | Common Voice 24 (MN) | Хан хурмаст уурлаж, Болдоггүй Бор өвгөнийг хор луугаараа ниргүүлэхээр явуулжээ. | ханхурамс дуурлас болдоггүй бороогүйн хоёр луугаар нэргүүлэхээр явуулжээ | 80.0% | 22.1% | 0/2/6 | |
| 6 | btsee_0006 | Common Voice 24 (MN) | Алив наашаа ороод ир гээд гэртээ оров. | алив наашаа ороодор гээд гэртэй оров | 42.9% | 8.1% | 0/1/2 | |
| 7 | btsee_0007 | Common Voice 24 (MN) | Өө өндөр дээдэс таны тухайд би баталж чадахгүй. | өө өндөр дээд таны тухайд би батал ч алдахгүй | 50.0% | 10.9% | 1/0/3 | |
| 8 | btsee_0008 | Common Voice 24 (MN) | Харин гурав дахь удаагаас эхлэн хүмүүсийг сонирхож эхлэв. | харин гурав дахь удаагаас эхлэн хүмүүсийг сонирхож эхлэв | 0.0% | 0.0% | 0/0/0 | |
| 9 | btsee_0009 | Common Voice 24 (MN) | Та нар очингуутаа шөл л өгч үз. | танар очингуутаа шөлөл өгчүүз | 85.7% | 10.0% | 0/3/3 | |
| 10 | btsee_0010 | Common Voice 24 (MN) | Ерөөсөө литр үйлдвэрээсээ салаагүй явсан юм чинь. | ерөөсөө л үйлдвэрээсээ салаагүй явсан юм чинь | 14.3% | 6.2% | 0/0/1 |
Methodology
How these numbers were produced.
Provider: Whisper large-v3 MN (Whisper large-v3 fine-tuned on Mongolian Common Voice 16.1 + FLEURS. Same author as the Common Voice 20 corpus in this benchmark, so its Common Voice scores should be read with train/test overlap in mind.). 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 warmestman/whisper-large-v3-mn-cv-fleurs (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 large-v3 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.