Bulgarian speech-to-text
GoodWe sent 30 Bulgarian test clips through this API. Whisper Large v3 made 8.4% word errors, about one word in 12; Parakeet made 11%. $0.002 per audio minute, word timestamps, SRT and VTT subtitles, files up to 100 MB.
Measured accuracy
| Model | Word error rate | Character error rate | Clips with no errors | Empty transcripts | Speed |
|---|---|---|---|---|---|
Whisper Large v3openai/whisper-large-v3 | 8.4% | 2.4% | 4 of 30 | 0 of 30 | 0.072× real time |
Parakeet TDT 0.6B v3nvidia/parakeet-tdt-0.6b-v3 | 11% | 3.3% | 4 of 30 | 0 of 30 | 0.053× real time |
Good. Usable with light review. Expect to correct a word here and there, mostly names and numbers. The clips are clean, single-speaker read speech, so treat this as a best case for your own recordings. How we measured, and all 83 languages.
What a transcript looks like
The clip with the median error rate from the test, reference text against what the API returned.
| Spoken | Но за птиците, които все още приличат на динозаври, има много неща. |
| Whisper Large v3 | Но за птиците, които все още приличат на джинозаври, има много неща. |
| Parakeet (its median clip) | Spoken: Сателитът беше изпратен в космоса от ракета. Учените използват телескопи в космоса, защото земната атмосфера изкривява част от светлината и изгледа ни. Сателитът беше изпратен в космоса от ракета, учените използват телескопи в космоса, защото земната атмосфера изкривява част от светлината и изгледане. |
Language detection
All 10 clips we sent without a language field were identified as Bulgarian. Auto-detection is safe here; sending language=bg still skips the detection step.
Which model for Bulgarian
Whisper Large v3 was more accurate on these clips (8.4% against 11%). Parakeet is faster, which matters mainly on long recordings. Both cost the same. Side by side.
Quickstart
The request is the OpenAI transcription request with a different base URL. language is optional.
curl https://api.inferenceapis.com/v1/audio/transcriptions \
-H "Authorization: Bearer $INFERENCE_API_KEY" \
-F "model=openai/whisper-large-v3" \
-F "language=bg" \
-F "file=@recording.mp3" \
-F "response_format=verbose_json"import os
from openai import OpenAI
client = OpenAI(base_url="https://api.inferenceapis.com/v1", api_key=os.environ["INFERENCE_API_KEY"])
with open("recording.mp3", "rb") as f:
r = client.audio.transcriptions.create(model="openai/whisper-large-v3", file=f, language="bg", response_format="verbose_json")
print(r.text)Subtitles and timestamps
response_format=srt or vtt returns a subtitle file with cues of up to two lines, broken at sentence ends and pauses. verbose_json with timestamp_granularities[]=word returns a start and end time for every word if you want to build your own.
Cost
$0.002 per minute of audio on either model: $0.12 for an hour, $12 for a hundred hours. Duration is measured from the file you upload. There is no daily cap on minutes and no subscription; enough to run your own Bulgarian test.
FAQ
How accurate is Whisper for Bulgarian?
On 30 clean read-speech clips it made 8.4% word errors, about one word in 12. Usable with light review. Expect to correct a word here and there, mostly names and numbers. Real recordings with noise, accents or several speakers will score worse.
Does language auto-detection work for Bulgarian?
In our test, yes: all 10 clips sent without a language were identified as Bulgarian. Sending language=bg still saves the detection step on every request.
Whisper or Parakeet for Bulgarian?
On the same clips Whisper Large v3 scored 8.4% and Parakeet 11%. Whisper was more accurate here; Parakeet is faster, which matters mainly for long recordings. Both cost $0.002 per minute.
Can it translate Bulgarian speech to English?
The API returns Bulgarian text. For English, transcribe first and translate the text with a chat model on the same key.
Related languages: Macedonian (16%) · Serbian (21%) · Russian (5.8%)