Struct SpeechState 

pub struct SpeechState { /* private fields */ }

The state of a speech input: captures audio from an AudioInput, feeds it to a SpeechRecognizer and tracks the transcript.

Render it with SpeechButton and SpeechWaveform, and subscribe to SpeechEvent to receive the text.

The recognizer is, in order: the one passed to Self::recognizer; else the platform’s SystemRecognizer, unless Self::system_fallback turned it off; else none, and the state is not available. The input defaults to the Microphone. Both defaults need the speech feature.

Implementations

impl SpeechState

pub fn new(_: &mut Context<'_, Self>) -> Self

Create a speech state with the default recognizer and input.

pub fn recognizer(self, recognizer: impl SpeechRecognizer) -> Self

Recognize speech with recognizer instead of the system’s.

pub fn input(self, input: impl AudioInput) -> Self

Capture audio from input instead of the microphone.

pub fn system_fallback(self, system_fallback: bool) -> Self

Whether to fall back to the platform’s recognizer when no Self::recognizer is set, default true.

On Windows the system recognizer dictates through Microsoft’s online service; turn this off when audio must not leave the application.

pub fn stop_timeout(self, timeout: Duration) -> Self

Set how long Self::stop waits for the recognizer’s final result before ending the session with the transcript so far, default 3 seconds.

pub fn has_recognizer(&self) -> bool

Whether a recognizer and an input are configured, so that speech input can work on this platform at all.

pub fn is_available(&self, cx: &App) -> bool

Whether a session can start now: Self::has_recognizer and the recognizer reports itself available.

pub fn status(&self) -> SpeechStatus

Where the state is in its session.

pub fn transcript(&self) -> SharedString

The transcript of the current or last session: every committed phrase followed by the current hypothesis.

pub fn levels(&self) -> impl ExactSizeIterator<Item = f32> + '_

Recent input levels in 0.0..=1.0, oldest first, one per 80 ms of audio. Peaks rise at once and fall back smoothly; background noise reads as 0.0.

pub fn start(&mut self, cx: &mut Context<'_, Self>)

Start a session. Does nothing while one is running.

When the recognizer or the input fails to start, emits SpeechEvent::Error and stays idle.

pub fn stop(&mut self, cx: &mut Context<'_, Self>)

Stop capturing and wait for the final result, which arrives as SpeechEvent::Final.

pub fn cancel(&mut self, cx: &mut Context<'_, Self>)

End the session at once and discard its transcript.

pub fn toggle(&mut self, cx: &mut Context<'_, Self>)

Start a session when idle, otherwise stop the running one.

Trait Implementations

impl EventEmitter<SpeechEvent> for SpeechState