Struct GeminiStt 

pub struct GeminiStt { /* private fields */ }

Available on crate feature audio only.

Gemini STT provider using generateContent with audio input.

Sends audio as inline data to the Gemini API and receives a text transcription. Supports language detection and optional prompting for specialized transcription tasks.

Example

use adk_audio::{GeminiStt, SttProvider, SttOptions, AudioFrame};

let stt = GeminiStt::from_env()?;
let transcript = stt.transcribe(&audio_frame, &SttOptions::default()).await?;
println!("Transcribed: {}", transcript.text);

Implementations

impl GeminiStt

pub fn from_env() -> Result<GeminiStt, AudioError>

Available on crate feature stt only.

Create from environment variable GEMINI_API_KEY or GOOGLE_API_KEY.

pub fn new(api_key: impl Into<String>) -> GeminiStt

Available on crate feature stt only.

Create with an explicit API key.

pub fn with_model(self, model: impl Into<String>) -> GeminiStt

Available on crate feature stt only.

Set the model to use for transcription.

pub fn with_prompt(self, prompt: impl Into<String>) -> GeminiStt

Available on crate feature stt only.

Set a custom transcription prompt.

The prompt is sent alongside the audio to guide the model’s output. For example: “Transcribe this audio in English with punctuation.”

Trait Implementations

impl SttProvider for GeminiStt

fn transcribe<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, audio: &'life1 AudioFrame, opts: &'life2 SttOptions, ) -> Pin<Box<dyn Future<Output = Result<Transcript, AudioError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, GeminiStt: 'async_trait,

Transcribe a single audio frame (batch mode).

fn transcribe_stream<'life0, 'life1, 'async_trait>( &'life0 self, _audio: Pin<Box<dyn Stream<Item = AudioFrame> + Send>>, _opts: &'life1 SttOptions, ) -> Pin<Box<dyn Future<Output = Result<Pin<Box<dyn Stream<Item = Result<Transcript, AudioError>> + Send>>, AudioError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, GeminiStt: 'async_trait,

Transcribe a stream of audio frames (streaming mode).