Swahili audio to text: what to expect
Swahili has more training data than most African languages in the speech model, but far less than English or French. Clear speech from one or two people gives a usable draft that needs a careful read. Names, places, numbers and fast casual speech need the most corrections. Noise, music and overlapping voices make the draft noticeably weaker.
Upload the recording as it is: MP3, M4A, WAV, OGG, OPUS, FLAC or AAC, a voice note saved from WhatsApp or Telegram, or a video file such as MP4 or MOV. You can also record from your microphone in the browser. Long files are split and processed in parallel, so a ten-minute recording usually takes a minute or two.
Spelling and word forms
Swahili uses the Latin alphabet with fairly regular spelling, which makes correcting by ear easier than in many languages. The main things to check are long words built from prefixes and suffixes, which the draft may split or join wrongly, and the sounds written ng', dh, th and gh.
Numbers may come out as words or digits, and Arabic-origin religious phrases may be spelled in different ways. Pick a style and keep it consistent during review.
Regional Swahili, Sheng and English
The model is closest to standard Swahili as heard in Tanzanian and Kenyan media. Congolese Swahili, coastal varieties and Ugandan speech will need more corrections.
In Nairobi, many young people speak Sheng, which mixes Swahili, English and other languages, and its slang changes quickly. The draft will often misspell Sheng words. Many professional conversations also switch to English. Keep Swahili selected when Swahili dominates; choose English for mostly English meetings.
What people record in Swahili
Uploads include research interviews and focus groups for public health and development projects, sermons and church events, radio segments, journalism interviews, lectures, and family voice messages. East African communities in the UK, the US and the Gulf use it for calls with relatives at home.
Exports and next steps
Export TXT or DOCX for notes and documents, PDF for a fixed copy, SRT or VTT for subtitles, and CSV with one row per line and its start time. Timestamps and speaker names can be switched on or off in each export. Speaker labels are available on paid plans or with a Minute Pack.
For video files, such as a YouTube sermon or a news clip, see the Swahili video to text page.