Dictation glossary

Speech-to-text

Speech-to-text is the process of converting spoken audio into written words using a speech recognition model. A microphone captures sound, software finds the speech in that signal, and a model maps acoustic patterns onto a sequence of words. The result can be a transcript, a caption, or text inserted into an app. Accuracy depends on audio quality, language, vocabulary, and the model.

Get WhisperJot Try the browser demo

In more detail

What is Speech-to-text?

What is speech to text in practice? The pipeline starts with a waveform: air pressure over time, sampled many thousands of times per second. Voice activity detection may drop silence so the recognizer spends compute on speech. The model then scores likely word sequences, often with a language component that prefers fluent phrases over random tokens. Decoding yields text, which downstream tools can punctuate, capitalize, or align to timestamps. Batch systems wait until a clip ends; streaming systems emit partial words while audio is still arriving.

People meet speech-to-text as captions, meeting notes, voicemail transcripts, and voice assistants. Dictation is the same conversion aimed at producing editable text in a writing workflow rather than an archive of a recording. The same model family can power both, but the product around it changes: a dictation app must place text in the focused field, handle commands such as new paragraph, and recover from noise without forcing you to replay a file. Word error rate is the usual accuracy yardstick; latency and privacy are the usual product tradeoffs.

For writers who speak

Why it matters for dictation

Dictation is speech-to-text aimed at the cursor. If recognition is slow, noisy, or wrong on names, you stop speaking and start editing, which erases the speed advantage over typing. If it is private, offline-capable, and accurate on your vocabulary, speaking becomes a practical way to draft email, notes, and documents without looking down at a keyboard. Understanding the term helps you compare captions, meeting transcripts, and live dictation as different jobs that share one conversion step.

In this product

How WhisperJot handles it

WhisperJot is a voice-dictation app for macOS, Windows, and Linux: press a hotkey, speak, and speech-to-text lands in the focused app. By default it uses the local engines, Jot Local and Jot Local Pro, which transcribe on your device and work offline after the first model download. Jot Cloud is an optional, opt-in engine that runs only while selected; audio is never stored on our servers.

Related terms

Keep reading in the glossary.

Voice dictation

Voice dictation is a way of writing in which you speak and software turns those words into editable text in an application.

Read Voice dictation →

On-device transcription

On-device transcription means speech recognition that runs on the computer or phone in front of you, using a model loaded in local memory.

Read On-device transcription →

Word error rate (WER)

Word error rate (WER) is a score of how many mistakes a speech recognizer made compared with a reference transcript.

Read WER →

Whisper

Whisper is an open-source family of automatic speech recognition models that convert recorded speech into text.

Read Whisper →

Questions

Straight answers.

What is speech to text?

Speech to text is software that turns spoken audio into written words. A microphone records sound, a model recognizes phones and words, and the system outputs a transcript or inserts text into an app. Captions, voicemail transcripts, and voice dictation all use this conversion. Quality depends on the microphone, background noise, language, and whether the vocabulary matches what you actually say.

How does speech-to-text software work?

Most systems slice audio into short frames, detect which frames contain speech, and run a neural recognizer that scores likely word sequences. A decoder then picks a transcript, sometimes with punctuation and timestamps. Cloud services send audio to a remote model; on-device engines load a model on your computer. Streaming setups emit partial text; batch setups wait until the utterance or file is complete.

Is speech-to-text the same as dictation?

Speech-to-text is the conversion from audio to words. Dictation is a workflow that uses that conversion so you can speak and have text appear where you are already writing. A meeting transcript can be speech-to-text without being dictation, because nobody is inserting words into a live document. The terms overlap in marketing, but the jobs differ: archive versus cursor.

One hotkey, any focused app.

Private local transcription by default, with an optional opt-in cloud engine.

Get WhisperJot — $99/year