Be realistic about Bashkir results
Bashkir is one of the languages the speech model knows least. Expect a rough draft with many errors, even on good audio. It works best on slow, clear speech from a single speaker close to the microphone. Conversations, songs, kurai music in the background or phone-quality audio will often produce text that needs rewriting rather than correcting.
Why use it at all? Correcting a timed draft while the player follows along is still quicker for many people than typing every word from scratch, especially for long interviews.
Upload the recording as it is: MP3, M4A, WAV, OGG, OPUS, FLAC, a voice message or a phone recording. You can also record from your microphone in the upload box. Nothing needs converting first.
Bashkir Cyrillic letters
Bashkir is written in Cyrillic with nine extra letters: ә, ө, ү, ғ, ҡ, ң, ҙ, ҫ and һ. These are the letters most likely to go wrong. The model may write the nearest Russian letter instead, such as к for ҡ or з for ҙ, so check them during review. Exports are UTF-8, so the special letters are kept in Word, CSV and subtitle files.
The transcript opens next to a player that highlights the current line. Click a line to hear it again, correct the word in place, and your changes save automatically.
Russian and Tatar in the same recording
Most Bashkir speakers in Bashkortostan also speak Russian, and many recordings switch between the two. Russian words in a Bashkir file may be kept in Russian spelling or bent toward Bashkir. If a recording is mostly Russian, choose Russian for a cleaner result.
Tatar is closely related and also in the language list. Speakers near the Tatarstan border may sound closer to Tatar; if the speaker considers their speech Tatar, use the Tatar page instead.
What people record in Bashkir
Typical files are interviews with elders for family or village history, folklore and song recordings collected by researchers, school and university language materials, and local radio or community events. A rough draft gives linguists and teachers something to annotate instead of an empty page. The audio repair switch, which reduces background noise, is worth trying on old field recordings.
Export TXT, DOCX, PDF, CSV, SRT or VTT, with timestamps and speaker names switched on or off per export. Speaker labels are available on paid plans or with a Minute Pack.
For video files, see the Bashkir video to text page.