Struct TranscriptionService 

pub struct TranscriptionService { /* private fields */ }

The main transcription service.

This service manages the entire transcription pipeline: capturing audio from the system’s default input device, chunking it into segments, sending those segments to an OpenAI-compatible transcription API, and emitting events with the results.

Examples

Basic usage

use vtt_rs::{TranscriptionService, Config};

let config = Config::default();
let api_key = std::env::var("OPENAI_API_KEY")?;

let mut service = TranscriptionService::new(config, api_key)?;
let (mut receiver, _stream) = service.start().await?;

// Process events...

With custom configuration

use vtt_rs::{TranscriptionService, Config};

let config = Config {
    chunk_duration_secs: 3,
    model: "whisper-1".to_string(),
    endpoint: "https://api.openai.com/v1/audio/transcriptions".to_string(),
    out_file: None,
};

let api_key = std::env::var("OPENAI_API_KEY")?;
let mut service = TranscriptionService::new(config, api_key)?;

Implementations

impl TranscriptionService

pub fn new(config: Config, api_key: String) -> Result<Self>

Creates a new transcription service with the specified configuration and API key.

This doesn’t start audio capture yet; call start to begin transcription.

Examples

use vtt_rs::{TranscriptionService, Config};

let config = Config::default();
let api_key = std::env::var("OPENAI_API_KEY")?;
let service = TranscriptionService::new(config, api_key)?;

Errors

Currently always succeeds, but returns Result for future extensibility.

pub fn new_no_api(config: Config) -> Result<Self>

Creates a new transcription service for remote APIs without an API key.

This is useful when connecting to a local OpenAI-compatible server that does not require authentication (e.g. a self-hosted MLX server running a Parakeet model).

pub fn new_on_device(config: Config) -> Result<Self>

Creates a transcription service configured for on-device inference.

pub async fn start( &mut self, ) -> Result<(UnboundedReceiver<TranscriptionEvent>, Stream)>

Starts the transcription service and returns a receiver for events.

This method begins capturing audio from the default input device and spawns background tasks to process and transcribe audio chunks. Events are delivered through the returned [UnboundedReceiver<TranscriptionEvent>].

The returned [Stream] must be kept alive for audio capture to continue. Dropping either the receiver or the stream will stop transcription.

Examples

use vtt_rs::{TranscriptionService, Config, TranscriptionEvent};

let config = Config::default();
let api_key = std::env::var("OPENAI_API_KEY")?;

let mut service = TranscriptionService::new(config, api_key)?;
let (mut receiver, _stream) = service.start().await?;

// Process transcription events
while let Some(event) = receiver.recv().await {
    match event {
        TranscriptionEvent::Transcription { chunk_id, text } => {
            println!("Chunk {}: {}", chunk_id, text);
        }
        TranscriptionEvent::Error { chunk_id, error } => {
            eprintln!("Error in chunk {}: {}", chunk_id, error);
        }
    }
}

Errors

Returns an error if:

• No default audio input device is available
• The audio device cannot be configured
• The configured output file cannot be opened or created

Panics

May panic if the audio system is not properly initialized (rare).