Hungarian speech-to-text API: 17% word error, measured

Measured accuracy

Model	Word error rate	Character error rate	Clips with no errors	Empty transcripts	Speed
Whisper Large v3 `openai/whisper-large-v3`	17%	5.8%	5 of 30	0 of 30	0.063× real time
Parakeet TDT 0.6B v3 `nvidia/parakeet-tdt-0.6b-v3`	16%	4.9%	4 of 30	0 of 30	0.043× real time

Fair. Gets the content across, but needs a human pass before it is published or quoted. 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	Az összekapcsoló elemeket elég hamar igénybe vették, hogy a pálya megőrizze a helyét. Ugyanakkor lassan rájöttek, hogy hatékonyabb lenne, ha ráhelyeznének egy vasdarabot.
Whisper Large v3	Az összekapcsolódó elemeket elég hamar ügénybe vették, hogy a pálya megőrizze a helyét. Ugyanakkor lassan ráérte, hogy hatékonyabb lenne, ha ráhelyeznének egy vastarabot.
Parakeet (its median clip)	Spoken: Késő vasárnap az Egyesült Államok elnöke, Donald Trump, egy, a sajtótitkár által közvetített nyilatkozatban kijelentette: az amerikai katonai egységek elhagyják Szíriát. Késő vasárnap az Egyesült Államok elnöke Donald Trump egy a sajtótitkár által közvetített nyilatkozatban kijelentette, hogy az Amerikai katonai egységek elhagyták szinriát.

Language detection

All 10 clips we sent without a language field were identified as Hungarian. Auto-detection is safe here; sending language=hu still skips the detection step.

Which model for Hungarian

The two models are within the noise of a 30-clip test (17% and 16%). Parakeet is faster; Whisper handles mixed-language audio and has automatic detection across 100 languages. Both cost the same, so try both on a sample of your own recordings. Side by side.

Quickstart

The request is the OpenAI transcription request with a different base URL. language is optional.

cURL

curl https://api.inferenceapis.com/v1/audio/transcriptions \
  -H "Authorization: Bearer $INFERENCE_API_KEY" \
  -F "model=openai/whisper-large-v3" \
  -F "language=hu" \
  -F "file=@recording.mp3" \
  -F "response_format=verbose_json"

Python · openai SDK

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="hu", 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 Hungarian test.

FAQ

How accurate is Whisper for Hungarian?

On 30 clean read-speech clips it made 17% word errors, about one word in 6. Gets the content across, but needs a human pass before it is published or quoted. Real recordings with noise, accents or several speakers will score worse.

Does language auto-detection work for Hungarian?

In our test, yes: all 10 clips sent without a language were identified as Hungarian. Sending language=hu still saves the detection step on every request.

Whisper or Parakeet for Hungarian?

On the same clips Whisper Large v3 scored 17% and Parakeet 16%. That is within the noise of a 30-clip test; Parakeet is faster, so try it first on long recordings. Both cost $0.002 per minute.

Can it translate Hungarian speech to English?

The API returns Hungarian text. For English, transcribe first and translate the text with a chat model on the same key.