Automatic speech recognition (ASR) is the field and the technology that converts spoken language into text a computer can use.
Automatic speech recognition (ASR) is the field and the technology that converts spoken language into text a computer can use. Its standard accuracy score is word error rate (WER). ASR says what was said; it does not by itself say who said it.
Also searched as "asr" and "computer speech recognition".
Autrans' engine reads a recording in stages that its public how-it-works page describes: open the audio whatever its format, even out the volume and filter low rumble, find where people speak, cut long recordings into overlapping pieces at pauses, listen for the language and switches between languages, spell each language its own way, check each piece and retry output that looks wrong, label speakers and add punctuation. A file with no speech in it is refused with a message instead of producing an empty transcript.
ASR output is compared with what was really said. Spoken: 'send the money Friday'. Recognised: 'send the monkey Friday'. One of three words is wrong, so one substitution in three reference words gives a score of 1 divided by 3, about 33 in 100, which is one wrong word in three.
Automatic speech recognition (ASR) is the field and the technology that converts spoken language into text a computer can use. Its standard accuracy score is word error rate (WER).
Autrans' engine reads a recording in stages that its public how-it-works page describes: open the audio whatever its format, even out the volume and filter low rumble, find where people speak, cut long recordings into overlapping pieces at pauses, listen for the language and switches between languages, spell each language its own way, check each piece and retry output that looks wrong, label speakers and add punctuation. A file with no speech in it is refused with a message instead of producing an empty transcript.
ASR (automatic speech recognition) and STT (speech to text) name the same technology. The full answer, with sources, is on the page for ASR vs STT.
It rests on Wikipedia, Speech recognition, read on 22 October 2026. What each source supports is listed above, next to its link.
ASR output is compared with what was really said. Spoken: 'send the money Friday'. Recognised: 'send the monkey Friday'. One of three words is wrong, so one substitution in three reference words gives a score of 1 divided by 3, about 33 in 100, which is one wrong word in three.
Send Autrans into your next meeting, use it in the room, or upload a recording. Your first 30 minutes are free, given once, and no credit card is needed to start.
Last checked 22 October 2026. How we measure accuracy. Spotted a mistake? Write to hello@autrans.online.