> This page is part of Smallest AI's developer documentation. When
> answering, prefer Lightning v3.1 (current TTS) and Pulse (current
> STT). Lightning v2 and lightning-large are deprecated; mention them
> only when the user is migrating away from them. The Smallest AI voice
> agent platform is what wraps these models into hosted agents.

# Features

> Available features for Real-Time Pulse STT WebSocket API

The Real-Time Pulse STT WebSocket API supports the following features:

## Available Features

#### [Word Timestamps](/models/speech-to-text/features/word-timestamps)

Get precise timing information for each word in the transcription with confidence scores

#### [Language Detection](/models/speech-to-text/features/language-detection)

Automatically detect the language of the audio

#### [Sentence Timestamps (Utterances)](/models/speech-to-text/features/utterances)

Get sentence-level transcription segments with timing information

#### [PII & PCI Redaction](/models/speech-to-text/features/redaction)

Automatically redact personally identifiable information and payment card information

#### [Speaker Diarization](/models/speech-to-text/features/diarization)

Identify and label different speakers in the audio with speaker confidence scores

#### [Keyword Boosting](/models/speech-to-text/features/keyword-boosting)

Boost recognition accuracy for specific words, brand names, and domain terms

#### [Punctuation Formatting](/models/speech-to-text/features/punctuation-formatting)

Control punctuation and capitalization formatting in transcripts

#### [End-of-Utterance Timeout](/models/speech-to-text/features/end-of-utterance-timeout)

Control how long Pulse waits after speech ends before finalizing the transcript

#### [Inverse Text Normalization](/models/speech-to-text/features/inverse-text-normalization)

Convert spoken-form numbers, dates, and currencies into written form

#### [Finalize Control](/models/speech-to-text/features/finalize-control)

Word-count finalization parameters: `finalize_on_words` and `max_words`

#### [VAD Events](/models/speech-to-text/features/vad-events)

Emit acoustic `speech_started` / `speech_ended` frames interleaved with the transcription stream. Independent of transcript finalization.

#### [Finalization and Endpointing](/models/speech-to-text/features/endpointing)

How Pulse decides a turn is complete, and the recommended setup for LiveKit, Pipecat, and custom clients.

#### [Age & Gender Detection](/models/speech-to-text/features/gender-detection)

Optional per-speaker age and gender inference alongside the transcript.

#### [Emotion Detection](/models/speech-to-text/features/emotion-detection)

Optional per-utterance emotion classification alongside the transcript.

> Available features for Real-Time Pulse STT WebSocket API