Glossary
Intent recognition
Intent recognition is the step that decides what a spoken request is asking for, such as setting a timer or sending a message, and which details the action needs. Voice assistants and voice agents rely on it before they act.
Updated October 17, 20263 cited facts, each traced to its source
Also called intent detection, natural language understanding, NLU.
Intent recognition (also called intent detection, natural language understanding or NLU) maps a request to the action it asks for. Amazon Web Services defines an intent in Amazon Lex as something that represents an action that the user wants to perform 1. The intent is the "what." The words you say are only the evidence for it.
How a request becomes an action
A voice system usually runs intent recognition after speech recognition has produced text. The steps run in order:
- The recognizer turns the audio into words.
- The intent step classifies the words, for example as a timer request or a message request.
- The system extracts the details the action needs, such as a duration or a recipient.
- It either carries out the action or asks a follow-up question when a detail is missing.
Take the request "Set a timer for ten minutes." The intent is a timer, and the duration is ten minutes. "Text Sam I'm running late" has a message intent, with a recipient and a body. If the recognizer mishears "Sam" as "Sand," the intent may still be right while the recipient is wrong, so the system has to check the details, not only the category.
Systems often try local matching first
Designs differ in where the matching happens. Home Assistant's Assist shows one common pattern. Home Assistant says Assist handles commands first by default, and only sends questions or commands it does not understand to the AI you have set up 2. A fixed set of known commands is matched locally, and a language model handles the rest.
Errors in the recognizer hurt intent accuracy. A 2024 study found that giving a language model the recognizer's lattice, not only its top transcript, helped bridge the gap to an oracle upper bound in spoken question answering and intent classification 3. The top transcript is one guess. The alternatives carry information that a single guess throws away.
Why it matters for dictation
Dictation rarely needs intent recognition for ordinary text. Its goal is the words themselves, not an action. Dictation tools still need a small intent step for a few spoken commands, such as spoken punctuation, and those commands are easy to misread.
The risk runs both ways. A command you meant as text can run as a command, and an instruction can end up typed into your document. When you judge a dictation tool, check how it separates the two, and whether you can turn off the commands you do not use. Our guide to speech to text AI covers how language models add a second pass on top of recognition, which is a different step from intent matching.
Related terms
- Voice assistant: the product that depends on intent recognition to act.
- Voice agent: a system that chains several intents and tool calls across a conversation.
- Automatic speech recognition: the step that produces the words intent recognition reads.
- Large language model: one way to classify intents without a fixed list of commands.
Guides on this topic
Evidence
Every number on this page, with where it comes from. Open a fact to see the exact words in its source.
- 1.
- 2.
- 3.
Related terms
- Speech to Text AI: Dictation, Assistants and Voice AgentsThe topic this term belongs to
- AI text detectorAn AI text detector is software that estimates whether a passage was written by a language model, usually from the words alone. Detectors make mistakes in both directions, so a score is weak evidence about how a text was made.
- Generative AIGenerative AI is the class of AI models that create new content, such as text, audio or images, from patterns in their training data. The NIST profile defines it as models that emulate the structure and characteristics of input data to generate derived synthetic content.
- Generative error correction (GER)Generative error correction (GER) passes a speech recognizer's output, often its N-best list of alternative transcripts, to a large language model, which rewrites the transcript to fix errors. It can recover words the recognizer never offered, but it can also change words that were already right.
- Large language model (LLM)A large language model (LLM) is a language model with a very large number of parameters, trained on huge amounts of text to predict and generate language. Google's glossary defines the minimum as a language model having a very high number of parameters.
- Machine learningMachine learning is a method in which a program learns its behavior from example data rather than from rules written by hand. Google's glossary defines it as a program or system that trains a model from input data.
- Prompt (AI)In AI, a prompt is the input you give a model to get a response, such as a question, an instruction or pasted text. Google's glossary defines it as any text entered as input to a large language model to condition the model to behave in a certain way.
- Speech language modelA speech language model (speech LLM) is a large language model that accepts audio as input, so it can process speech directly instead of working only on a text transcript. The 2023 SALMONN model is one example that lets an LLM process general audio inputs.
- Speech-to-speech modelA speech-to-speech model takes audio in and produces audio out with one model, without writing the conversation out as text in between. Most voice systems instead chain speech recognition, a language model and speech synthesis.
- Text to speech (TTS)Text to speech (TTS), also called speech synthesis, is software that converts written text into spoken audio. Modern systems use neural networks, and they can run on the device or in the cloud.
- Turn detectionTurn detection is the step that decides when a speaker has finished a turn, so a voice system knows when to respond. A pause can be a hesitation rather than an ending, so the decision trades speed against cutting the speaker off.
- Voice agentA voice agent is a voice system that listens, reasons with a language model, uses tools, and speaks back across several turns to finish a task, such as checking an order and rescheduling it. Benchmarks show that agents still fail many tasks on realistic audio.
- Voice assistantA voice assistant is software that listens for a spoken request, works out what you want, and then acts or answers, such as setting a timer or turning off a light. Dictation is different: it turns your speech into text for you to edit.
Pages that use this term
- Speech to Text AI: Dictation, Assistants and Voice AgentsTopic
- Voice Assistants vs Dictation: Siri, Alexa, GeminiGuide
- Home Assistant Voice: Speech to Text and Voice CommandsGuide
- Can Home Assistant use ChatGPT or a local LLM for voice?Question
- What is the best speech to text for Home Assistant?Question
- What voice commands does Home Assistant understand?Question
- Why do voice assistants misunderstand you?Question
- Voice agentGlossary
- Voice assistantGlossary