Speech-to-text
Technology that converts spoken audio into written text.
Speech-to-text, sometimes called ASR for automatic speech recognition, does exactly what the name says: audio in, words out. It is the layer underneath meeting notes, video subtitles, podcast transcripts, voice typing and every voice assistant. Modern versions handle multiple languages, tell speakers apart and can be given a list of names or jargon in advance to improve accuracy.
Two flavours exist. Batch models take a finished recording and process it faster than real time, often thirty times faster or more. Streaming models transcribe as you speak, which is harder because they cannot look ahead at what comes next. Prices have collapsed to fractions of a cent per minute, and the accuracy differences between leading providers are now small enough that convenience often matters more.
-
Google's New Transcription Model Cleans Up Your Ums and Your Mind Changes
-
IBM's Granite 4.2 Is Free, Small Enough to Run Locally, and Built to Think First
-
These AI Glasses Deliberately Have No Camera. That Turns Out to Be the Whole Point
-
OpenAI's new transcription models are faster and 25 percent cheaper, but still not the most accurate
-
Microsoft Is Quietly Swapping OpenAI and Anthropic Out of Copilot to Cut Costs