Turn a box of cassette tapes into a searchable Yoruba archive
Mansa transcribes oral recordings across African languages and keeps the original wording intact, so preservation work doesn't depend on how many hours one researcher has free this month.
Category: Research & Archives · Languages: 30+ African languages · Product: Mansa Transcribe
The challenge
A lot of what makes up a community's linguistic memory only exists as sound: an elder's folktale on a cassette, a griot's praise-song recorded on a phone, a linguist's decades-old field tapes sitting in a box. None of it is searchable. None of it can be cited, indexed, or cross-referenced until someone sits down and transcribes it by hand, which is slow, expensive, and rare for languages most transcription services don't support well. Every year that passes without documentation is a year closer to losing a dialect, a story, or a way of speaking for good.
How Mansa helps
Mansa Transcribe is a standalone tool. You select the source language, then either upload an audio file or record directly in the app, and hit Transcribe. It returns the speech as text in that same source language, preserving the speaker's exact wording.
In practice
- 1
Select the source language
Choose the language being spoken in the recording, for example Yoruba.
- 2
Upload or record
Upload an existing audio file, a cassette digitization, a phone recording, a field tape, or record live if you're sitting with the elder or interviewee in person.
- 3
Hit Transcribe
Mansa returns the full transcript as text in the source language, with the speaker's original wording preserved.
If you need it in another language too
To get an English version of the transcript, take the output and run it through Mansa Translate separately: choose Yoruba as the source, English as the target, and paste the transcript in.
Tips
- Upload the highest quality audio you have. Background noise and low bitrate audio reduce transcription accuracy more than dialect or accent does.
- If a recording mixes two languages, select whichever language dominates the recording as the source language, since Transcribe works from a single selected source language.