Whisper large-v3-turbo — Mongolian Speech-to-Text Benchmark
Real results from the Whisper large-v3-turbo 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-turbo list pricing for batch transcription. No discounts, no negotiated rates applied — the raw per-minute rate.
Pricing source: Whisper large-v3-turbo public pricing. Duudlaga Flow is shown for context only — this page isolates Whisper large-v3-turbo 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-turbo 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 | 100.3% | 56.2% | -0.3% | 0.86× | 330.6s | 382.6s |
| Shunya Labs Mongolian Speech | Hugging Face → | 60 | 60/60 | 99.4% | 47.7% | 0.6% | 1.01× | 645.1s | 640.2s |
| Common Voice 20 (MN) | Hugging Face → | 54 | 54/54 | 101% | 58.1% | -1.0% | 0.55× | 269.8s | 489.7s |
| Modern Voice | — | 92 | 92/92 | 96.8% | 55.6% | 3.2% | 1.18× | 603.8s | 513.8s |
WER vs speed — per sample
Each dot is one audio sample. The sweet spot is the bottom-left (low error, fast). Whisper large-v3-turbo 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-turbo got wrong, in red — no strikethrough/swap gymnastics, just the mistakes.
| # | Audio | Sample | Dataset | Expected (ground truth) | Whisper large-v3-turbo result | WER | CER | I/D/S |
|---|---|---|---|---|---|---|---|---|
| 1 | btsee_0001 | Common Voice 24 (MN) | Гэхдээ амьсгал хураахаасаа өмнө танд мэдэж байгаагаа хэлье. | Дээ амсхалд бараа бээхэээ тэндэ мэдэжэээгээээ хэлэээ. | 100.0% | 56.9% | 0/1/7 | |
| 2 | btsee_0002 | Common Voice 24 (MN) | Надад заяасан аз жаргал гэдэг ердөө гуравхан сарын хугацаатай байсан гэж үү? | нады заясан аджырғалгеттэгэртэ құрыққын сарай құтсаттэ бәсэнгэ чуэл | 100.0% | 54.7% | 0/4/8 | |
| 3 | btsee_0003 | Common Voice 24 (MN) | Одоо бид өөрсдөө өвчин эмгэгээсээ салахыг хичээцгээе. | Одо дэ дүр стөл чин рэгэсэсэлэй чэцгэй | 100.0% | 51.9% | 0/0/7 | |
| 4 | btsee_0004 | Common Voice 24 (MN) | Би бол голдуу хээрээр гэр, хэцээр дэр хийж явдаг хүн. | Би боз қолтүү хэрэр гэр цэр дэр хий жабдэг хүн | 60.0% | 29.4% | 0/0/6 | |
| 5 | btsee_0005 | Common Voice 24 (MN) | Хан хурмаст уурлаж, Болдоггүй Бор өвгөнийг хор луугаараа ниргүүлэхээр явуулжээ. | Hann formast orðlas boltgu, borðu hún hærð, þógar, neyrgur fyrir og sér. | 120.0% | 92.2% | 2/0/10 | |
| 6 | btsee_0006 | Common Voice 24 (MN) | Алив наашаа ороод ир гээд гэртээ оров. | Эсфина þessi орður, get gist í orð. | 100.0% | 83.8% | 0/0/7 | |
| 7 | btsee_0007 | Common Voice 24 (MN) | Өө өндөр дээдэс таны тухайд би баталж чадахгүй. | Уүндөрдээд тэний түхээд би бодал чадагүй. | 87.5% | 37.0% | 0/2/5 | |
| 8 | btsee_0008 | Common Voice 24 (MN) | Харин гурав дахь удаагаас эхлэн хүмүүсийг сонирхож эхлэв. | хараан горубтойг үудагаас сай кайлиң хүнүүсыйг сан рэх бэг келл ў. | 137.5% | 62.5% | 3/0/8 | |
| 9 | btsee_0009 | Common Voice 24 (MN) | Та нар очингуутаа шөл л өгч үз. | Танар отжунгута шүлэл ўүтжүт. | 100.0% | 43.3% | 0/3/4 | |
| 10 | btsee_0010 | Common Voice 24 (MN) | Ерөөсөө литр үйлдвэрээсээ салаагүй явсан юм чинь. | Ёрөз үлтурэсэй салагы афсэймчэн. | 100.0% | 58.3% | 0/3/4 |
Methodology
How these numbers were produced.
Provider: Whisper large-v3-turbo (Whisper large-v3-turbo open weights (4 decoder layers) run locally — the speed-optimised distillation of large-v3.).
Endpoint local (transformers). Language mn. Operating point openai/whisper-large-v3-turbo (language=mn, task=transcribe) 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-turbo 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.