settings: Add ORT thread count setting with auto-tune benchmark.

src-tauri/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "handy"
-version = "0.8.0"
+version = "0.7.11"
 description = "Handy"
 authors = ["cjpais"]
 edition = "2021"
@@ -68,8 +68,7 @@ chrono = "0.4"
 rusqlite = { version = "0.37", features = ["bundled"] }
 tar = "0.4.44"
 flate2 = "1.0"
-sha2 = "0.10"
-transcribe-rs = { version = "0.3.2", features = ["whisper-cpp", "onnx"] }
+transcribe-rs = { git = "https://github.com/andrewleech/transcribe-rs", branch = "feat/ort-thread-count", features = ["whisper-cpp", "onnx"] }
 handy-keys = "0.2.4"
 ferrous-opencc = "0.2.3"
 clap = { version = "4", features = ["derive"] }
@@ -88,7 +87,7 @@ tauri-plugin-single-instance = "2.3.2"
 tauri-plugin-updater = "2.10.0"
 
 [target.'cfg(windows)'.dependencies]
-transcribe-rs = { version = "0.3.2", features = ["whisper-vulkan", "ort-directml"] }
+transcribe-rs = { git = "https://github.com/andrewleech/transcribe-rs", branch = "feat/ort-thread-count", features = ["ort-directml"] }
 windows = { version = "0.61.3", features = [
   "Win32_Media_Audio_Endpoints",
   "Win32_System_Com_StructuredStorage",
@@ -100,12 +99,10 @@ winreg = "0.55"
 
 [target.'cfg(target_os = "macos")'.dependencies]
 tauri-nspanel = { git = "https://github.com/ahkohd/tauri-nspanel", branch = "v2.1" }
-transcribe-rs = { version = "0.3.2", features = ["whisper-metal"] }
 
 [target.'cfg(target_os = "linux")'.dependencies]
 gtk-layer-shell = { version = "0.8", features = ["v0_6"] }
 gtk = "0.18"
-transcribe-rs = { version = "0.3.2", features = ["whisper-vulkan"] }
 
 [patch.crates-io]
 tauri-runtime = { git = "https://github.com/cjpais/tauri.git", branch = "handy-2.10.2" }

src-tauri/src/audio_toolkit/audio/mod.rs

@@ -6,7 +6,7 @@ mod utils;
 mod visualizer;
 
 pub use device::{list_input_devices, list_output_devices, CpalDeviceInfo};
-pub use recorder::{is_microphone_access_denied, is_no_input_device_error, AudioRecorder};
+pub use recorder::{is_microphone_access_denied, AudioRecorder};
 pub use resampler::FrameResampler;
-pub use utils::{read_wav_samples, save_wav_file, verify_wav_file};
+pub use utils::{load_wav_samples, save_wav_file, wav_duration_secs};
 pub use visualizer::AudioVisualiser;

src-tauri/src/audio_toolkit/audio/utils.rs

@@ -3,32 +3,34 @@ use hound::{WavReader, WavSpec, WavWriter};
 use log::debug;
 use std::path::Path;
 
-/// Read a WAV file and return normalised f32 samples.
-pub fn read_wav_samples<P: AsRef<Path>>(file_path: P) -> Result<Vec<f32>> {
-    let reader = WavReader::open(file_path.as_ref())?;
-    let samples = reader
-        .into_samples::<i16>()
+/// Load a 16-bit PCM WAV file and return samples as f32 in [-1, 1].
+/// Only mono 16-bit PCM files are supported (the format Handy writes).
+pub fn load_wav_samples(path: &Path) -> Result<Vec<f32>> {
+    let mut reader = WavReader::open(path)?;
+    let spec = reader.spec();
+    anyhow::ensure!(
+        spec.bits_per_sample == 16 && spec.sample_format == hound::SampleFormat::Int,
+        "load_wav_samples: expected 16-bit PCM, got {:?}",
+        spec
+    );
+    let samples: Result<Vec<f32>, _> = reader
+        .samples::<i16>()
         .map(|s| s.map(|v| v as f32 / i16::MAX as f32))
-        .collect::<Result<Vec<f32>, _>>()?;
-    Ok(samples)
+        .collect();
+    Ok(samples?)
 }
 
-/// Verify a WAV file by reading it back and checking the sample count.
-pub fn verify_wav_file<P: AsRef<Path>>(file_path: P, expected_samples: usize) -> Result<()> {
-    let reader = WavReader::open(file_path.as_ref())?;
-    let actual_samples = reader.len() as usize;
-    if actual_samples != expected_samples {
-        anyhow::bail!(
-            "WAV sample count mismatch: expected {}, got {}",
-            expected_samples,
-            actual_samples
-        );
-    }
-    Ok(())
+/// Return the duration of a WAV file in seconds by reading its header only.
+/// Does not decode sample data.
+pub fn wav_duration_secs(path: &Path) -> Result<f32> {
+    let reader = WavReader::open(path)?;
+    let spec = reader.spec();
+    let duration = reader.duration();
+    Ok(duration as f32 / spec.sample_rate as f32)
 }
 
 /// Save audio samples as a WAV file
-pub fn save_wav_file<P: AsRef<Path>>(file_path: P, samples: &[f32]) -> Result<()> {
+pub async fn save_wav_file<P: AsRef<Path>>(file_path: P, samples: &[f32]) -> Result<()> {
     let spec = WavSpec {
         channels: 1,
         sample_rate: 16000,

src-tauri/src/audio_toolkit/mod.rs

@@ -5,8 +5,8 @@ pub mod utils;
 pub mod vad;
 
 pub use audio::{
-    is_microphone_access_denied, is_no_input_device_error, list_input_devices, list_output_devices,
-    read_wav_samples, save_wav_file, verify_wav_file, AudioRecorder, CpalDeviceInfo,
+    is_microphone_access_denied, list_input_devices, list_output_devices, load_wav_samples,
+    save_wav_file, wav_duration_secs, AudioRecorder, CpalDeviceInfo,
 };
 pub use text::{apply_custom_words, filter_transcription_output};
 pub use utils::get_cpal_host;