lobo/stt
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

switch to actual sample rate

Browse source
This commit is contained in:
Emil Lerch 2025-10-06 08:22:30 -07:00
parent 6559bf86a3
commit 2b2f50491d
Signed by: lobo
GPG key ID: A7B62D657EF764F8
1 changed files with 15 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
src/stt.zig
View file

@ -418,6 +418,12 @@ pub const AlsaCapture = struct {
err = c.snd_pcm_hw_params_set_rate_near(self.pcm_handle, hw_params, &actual_rate, null);
if (err < 0) return Error.SetSampleRateError;
// Update sample rate if hardware doesn't support requested rate
if (actual_rate != self.sample_rate) {
std.log.info("Hardware doesn't support {}Hz, using {}Hz", .{ self.sample_rate, actual_rate });
self.sample_rate = actual_rate;
}
// Set buffer size
var actual_buffer_size: c.snd_pcm_uframes_t = self.buffer_size;
err = c.snd_pcm_hw_params_set_buffer_size_near(self.pcm_handle, hw_params, &actual_buffer_size);
@ -602,7 +608,7 @@ pub const Session = struct {
}
// Initialize Vosk audio buffer (larger buffer for processing)
const vosk_audio_buffer = AudioBuffer.init(allocator, options.sample_rate * 2) catch {
const vosk_audio_buffer = AudioBuffer.init(allocator, alsa_capture.sample_rate * 2) catch {
const error_info = ErrorInfo.init(Error.OutOfMemory, "Failed to allocate Vosk audio buffer during initialization");
options.event_handler.onDetailedError(error_info);
return Error.OutOfMemory;
@ -946,8 +952,15 @@ pub const Session = struct {
return Error.InvalidState;
}
// Resample to 16kHz if needed
var resampled_buffer: [4096]i16 = undefined;
const final_audio = if (self.alsa_capture.?.sample_rate != 16000) blk: {
const resampled_count = AudioConverter.resample(audio_data, &resampled_buffer, self.alsa_capture.?.sample_rate, 16000);
break :blk resampled_buffer[0..resampled_count];
} else audio_data;
// Convert i16 samples to bytes for Vosk
const audio_bytes = std.mem.sliceAsBytes(audio_data);
const audio_bytes = std.mem.sliceAsBytes(final_audio);
// Feed audio to Vosk recognizer
const accept_result = c.vosk_recognizer_accept_waveform(self.vosk_recognizer, audio_bytes.ptr, @intCast(audio_bytes.len));