Guides › Transcribe Yoruba, Hausa and Pidgin audio

How to transcribe Yoruba, Hausa and Pidgin audio to text

By Lawal Ayo, LAS G Concept · Updated 15 October 2026

Interviews, church recordings, voice notes, market-research calls and radio clips are often in Yoruba, Hausa or Nigerian Pidgin, and most transcription tools handle them badly. LAS Converter's Audio → Text tool lets you pick the spoken language, supports English, Pidgin, Yoruba and Hausa, and can translate the result into English. This guide explains how to use it and how to get the best accuracy.

Speech to text is a Premium feature. Please read the notes on accuracy below: machine transcription of these languages is useful, but it is never perfect.

Step by step

Open the converter, log in with a Premium account, and choose Audio → Text.
Drop an audio or video file: MP3, WAV, M4A, AAC, FLAC, OGG, MP4, MOV, MKV or WebM all work.
Under Spoken Audio Source Language, choose English, Nigerian Pidgin, Yoruba or Hausa. Choosing the right language matters more than any other setting. Auto-detect is available, but is less reliable on short clips.
Optionally choose a Target Translation Language, for example English, to get an English version of Yoruba, Hausa or Pidgin speech.
Start the transcription and wait. Long files are cut into pieces and processed one after another, with timestamps put back in order.
Review the text, then export as SRT, VTT, ASS or plain TXT.

How it works behind the scenes

The tool can run a speech model on your own device, or use a larger cloud model. On-device models are Whisper Tiny (about 45 MB download), Base (about 80 MB, the recommended default) and Small (about 250 MB, slower but better for Yoruba, Hausa and noisy audio). They download once and then run in your browser, so the audio stays on your device.

For Pidgin, Yoruba and Hausa, Premium accounts can also use the server's larger AI model, which handles these languages better. In that case the audio is sent securely to our server and processed by our AI provider (Groq) only to produce your transcript. See the Privacy Policy for details. Each piece sent to the server is up to 24 MB.

Nigerian Pidgin is transcribed as English text, because Pidgin is written using mostly English spelling.

Get better accuracy

Clean audio wins. Record close to the speaker and avoid music or generator noise. A good voice note beats a noisy studio recording.
One language at a time. If a speaker mixes Yoruba and English, results are weaker. Split the file if you can.
Use the larger model (Small on-device, or the server model) for Yoruba and Hausa.
Convert to mono MP3 first if your file is huge. See MP4 to MP3.
Always proofread. Names, tone marks and local expressions are the most common mistakes. Do not publish a transcript of a legal, medical or official recording without a human check.

Igbo and other languages

The language list also includes Igbo (marked experimental), French, Spanish, German, Arabic, Chinese, Japanese, Portuguese, Italian and Russian. Expect weaker results for Igbo than for English.

What to do with the transcript

Plain text is good for notes and minutes. SRT and VTT files include timings, so you can use them as subtitles; see how to make SRT or VTT subtitles.

Frequently asked questions

Is speech to text free?

No, it is a Premium feature (₦1,500 a month or ₦12,000 a year).

Which languages are supported?

English, Nigerian Pidgin, Yoruba and Hausa, plus Igbo (experimental) and several world languages.

Is my audio kept?

On-device transcription never leaves your device. For server transcription the audio is processed to produce the transcript and is not kept after the job finishes.

How accurate is it?

It depends on audio quality and accent. English is best, Yoruba and Hausa are good on clear speech, Igbo is experimental. Always proofread.

Try it

Use the speech to text page to learn more, or go straight to the app.

Open the converter Create a free account