What Is Speech-to-Text? How AI Transcribes Spoken Audio
The reverse of text-to-speech: converting spoken audio into written text, powering everything from voice assistants to automatic meeting transcripts.
Turning spoken audio into written text
Speech-to-text (also called automatic speech recognition, or ASR) converts spoken audio into written text, handling variation in accent, pace, background noise, and overlapping speakers well enough to produce a usable, accurate transcript rather than a rough approximation.
Why this is a genuinely hard problem
Unlike written text, spoken audio has no clear boundaries between words, varies enormously between speakers, and is often degraded by background noise or a poor microphone. Modern systems handle this with models trained on massive amounts of diverse audio, learning to be robust to exactly that kind of real-world variation.
Where it fits in modern AI products
Speech-to-text is the entry point for any voice-driven AI interaction — a voice assistant, a phone-based support agent, an automatic meeting transcript — converting spoken input into text a language model can then actually process and respond to.
Frequently Asked Questions
What is speech-to-text?
AI technology that converts spoken audio into written text, handling variation in accent, pace, and background noise well enough to produce an accurate, usable transcript.
What is speech-to-text used for?
It's the entry point for voice-driven AI interactions — voice assistants, phone-based support agents, and automatic meeting transcription all rely on it to convert spoken input into text a model can process.