# Speech-to-Text (Speech Recognition)

> AI that listens to speech and writes the words down.

**Field:** Artificial Intelligence  
**Category:** Multimodal  
**Also known as:** Speech-to-Text, Speech Recognition, Automatic Speech Recognition, ASR, Transcription

## Think of it like

It is like a friend who writes down every word you say. You talk, they type.

## Definition

Speech-to-text is AI that turns spoken words into written text. The model listens to the sound, cuts it into tiny slices, and works out which words were most likely said. Today's models learn from many thousands of hours of speech, so they cope with accents and some noise. Many also add commas and full stops, mark who is speaking, and give each word a time. Quality is often scored by word error rate, the share of words it gets wrong.

## Example

Jack records a one-hour team meeting. In a few minutes, speech-to-text turns it into a written record, with each speaker marked. Then he searches it for the word 'budget'.

## Why it matters

Spoken words become text you can search and quote. It also opens up sound and video to people who cannot hear.

## Common mistake

Trusting names and numbers without a check. These are the words a model most often gets wrong.

## Three principles

- **Sound in.** Words out.
- **Trained on speech.** Thousands of hours.
- **Word error rate.** The usual score.

## Related terms

- [Text-to-Speech](https://learn.benibix.com/term/ai/text-to-speech)
- [Voice Agent](https://learn.benibix.com/term/ai/voice-agent)
- [Multimodal Model](https://learn.benibix.com/term/ai/multimodal-model)
- [Transformer](https://learn.benibix.com/term/ai/transformer)
- [AI Meeting Assistant](https://learn.benibix.com/term/ai/ai-meeting-assistant)
- [Whisper](https://learn.benibix.com/term/ai/whisper)
- [Speaker Diarization](https://learn.benibix.com/term/ai/speaker-diarization)

## Told apart from

- [Speech-to-Text vs Text-to-Speech](https://learn.benibix.com/compare/speech-to-text-vs-text-to-speech): People often mix up speech-to-text and text-to-speech, because the names are mirror images. They run in opposite directions. Speech-to-text listens to spoken words and writes them down. Text-to-speech takes written words and reads them out loud. A voice agent uses both: one to hear you, and one to answer.
- [Speaker Diarization vs Speech-to-Text](https://learn.benibix.com/compare/speaker-diarization-vs-speech-to-text): People think speech-to-text already says who spoke. It does not. Speech-to-text writes down the words. Speaker diarization splits the sound by voice and labels each part Speaker 1, Speaker 2 and so on. A meeting script needs both.

## Taught in

- [Voice AI and AI Companions](https://learn.benibix.com/learn/ai-use-case/voice-ai-and-ai-companions): AI Assistants and Chatbots, 2/13

![A sound wave of someone speaking goes into a speech model and comes out as written words with times beside them.](https://learn.benibix.com/diagrams/mm-speech-to-text.svg)

---

Source: https://learn.benibix.com/term/ai/speech-to-text
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-17
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
