From fa26a5511d5e14f97551bdd1d427ed004b7acd80 Mon Sep 17 00:00:00 2001 From: Marcelo Fernandez Date: Fri, 19 Oct 2018 16:32:03 -0300 Subject: [PATCH] Fix possible audio input buffer issues --- drivers/coreaudio/audio_driver_coreaudio.cpp | 5 ++--- drivers/pulseaudio/audio_driver_pulseaudio.cpp | 18 ++++++++---------- drivers/pulseaudio/audio_driver_pulseaudio.h | 1 - drivers/wasapi/audio_driver_wasapi.cpp | 5 +---- servers/audio/audio_stream.cpp | 16 +++++++++++++--- servers/audio/audio_stream.h | 2 -- servers/audio_server.cpp | 8 ++++++++ servers/audio_server.h | 1 + 8 files changed, 33 insertions(+), 23 deletions(-) diff --git a/drivers/coreaudio/audio_driver_coreaudio.cpp b/drivers/coreaudio/audio_driver_coreaudio.cpp index aaec63d7ffe..850b90d59b9 100644 --- a/drivers/coreaudio/audio_driver_coreaudio.cpp +++ b/drivers/coreaudio/audio_driver_coreaudio.cpp @@ -145,9 +145,6 @@ Error AudioDriverCoreAudio::init() { unsigned int buffer_size = buffer_frames * channels; samples_in.resize(buffer_size); input_buf.resize(buffer_size); - input_buffer.resize(buffer_size * 8); - input_position = 0; - input_size = 0; print_verbose("CoreAudio: detected " + itos(channels) + " channels"); print_verbose("CoreAudio: audio buffer frames: " + itos(buffer_frames) + " calculated latency: " + itos(buffer_frames * 1000 / mix_rate) + "ms"); @@ -487,6 +484,8 @@ void AudioDriverCoreAudio::capture_finish() { Error AudioDriverCoreAudio::capture_start() { + input_buffer_init(buffer_frames); + OSStatus result = AudioOutputUnitStart(input_unit); if (result != noErr) { ERR_PRINTS("AudioOutputUnitStart failed, code: " + itos(result)); diff --git a/drivers/pulseaudio/audio_driver_pulseaudio.cpp b/drivers/pulseaudio/audio_driver_pulseaudio.cpp index 9c02549e396..d78316945f3 100644 --- a/drivers/pulseaudio/audio_driver_pulseaudio.cpp +++ b/drivers/pulseaudio/audio_driver_pulseaudio.cpp @@ -613,20 +613,18 @@ Error AudioDriverPulseAudio::capture_init_device() { break; } - print_verbose("PulseAudio: detected " + itos(pa_rec_map.channels) + " input channels"); - pa_sample_spec spec; spec.format = PA_SAMPLE_S16LE; spec.channels = pa_rec_map.channels; spec.rate = mix_rate; - int latency = 30; - input_buffer_frames = closest_power_of_2(latency * mix_rate / 1000); - int buffer_size = input_buffer_frames * spec.channels; + int input_latency = 30; + int input_buffer_frames = closest_power_of_2(input_latency * mix_rate / 1000); + int input_buffer_size = input_buffer_frames * spec.channels; pa_buffer_attr attr; - attr.fragsize = buffer_size * sizeof(int16_t); + attr.fragsize = input_buffer_size * sizeof(int16_t); pa_rec_str = pa_stream_new(pa_ctx, "Record", &spec, &pa_rec_map); if (pa_rec_str == NULL) { @@ -642,9 +640,10 @@ Error AudioDriverPulseAudio::capture_init_device() { ERR_FAIL_V(ERR_CANT_OPEN); } - input_buffer.resize(input_buffer_frames * 8); - input_position = 0; - input_size = 0; + input_buffer_init(input_buffer_frames); + + print_verbose("PulseAudio: detected " + itos(pa_rec_map.channels) + " input channels"); + print_verbose("PulseAudio: input buffer frames: " + itos(input_buffer_frames) + " calculated latency: " + itos(input_buffer_frames * 1000 / mix_rate) + "ms"); return OK; } @@ -760,7 +759,6 @@ AudioDriverPulseAudio::AudioDriverPulseAudio() { mix_rate = 0; buffer_frames = 0; - input_buffer_frames = 0; pa_buffer_size = 0; channels = 0; pa_ready = 0; diff --git a/drivers/pulseaudio/audio_driver_pulseaudio.h b/drivers/pulseaudio/audio_driver_pulseaudio.h index f8358a452bc..d8bab841ffd 100644 --- a/drivers/pulseaudio/audio_driver_pulseaudio.h +++ b/drivers/pulseaudio/audio_driver_pulseaudio.h @@ -64,7 +64,6 @@ class AudioDriverPulseAudio : public AudioDriver { unsigned int mix_rate; unsigned int buffer_frames; - unsigned int input_buffer_frames; unsigned int pa_buffer_size; int channels; int pa_ready; diff --git a/drivers/wasapi/audio_driver_wasapi.cpp b/drivers/wasapi/audio_driver_wasapi.cpp index a91e41b0086..8665f701b19 100644 --- a/drivers/wasapi/audio_driver_wasapi.cpp +++ b/drivers/wasapi/audio_driver_wasapi.cpp @@ -336,10 +336,7 @@ Error AudioDriverWASAPI::init_capture_device(bool reinit) { HRESULT hr = audio_input.audio_client->GetBufferSize(&max_frames); ERR_FAIL_COND_V(hr != S_OK, ERR_CANT_OPEN); - // Set the buffer size - input_buffer.resize(max_frames * CAPTURE_BUFFER_CHANNELS); - input_position = 0; - input_size = 0; + input_buffer_init(max_frames); return OK; } diff --git a/servers/audio/audio_stream.cpp b/servers/audio/audio_stream.cpp index 7de0695e8c8..02a0bed9649 100644 --- a/servers/audio/audio_stream.cpp +++ b/servers/audio/audio_stream.cpp @@ -136,16 +136,20 @@ void AudioStreamPlaybackMicrophone::_mix_internal(AudioFrame *p_buffer, int p_fr Vector buf = AudioDriver::get_singleton()->get_input_buffer(); unsigned int input_size = AudioDriver::get_singleton()->get_input_size(); + int mix_rate = AudioDriver::get_singleton()->get_mix_rate(); + int playback_delay = MIN(((50 * mix_rate) / 1000) * 2, buf.size() >> 1); +#ifdef DEBUG_ENABLED + unsigned int input_position = AudioDriver::get_singleton()->get_input_position(); +#endif - // p_frames is multiplied by two since an AudioFrame is stereo - if ((p_frames + MICROPHONE_PLAYBACK_DELAY * 2) > input_size) { + if (playback_delay > input_size) { for (int i = 0; i < p_frames; i++) { p_buffer[i] = AudioFrame(0.0f, 0.0f); } input_ofs = 0; } else { for (int i = 0; i < p_frames; i++) { - if (input_size >= input_ofs) { + if (input_size > input_ofs) { float l = (buf[input_ofs++] >> 16) / 32768.f; if (input_ofs >= buf.size()) { input_ofs = 0; @@ -162,6 +166,12 @@ void AudioStreamPlaybackMicrophone::_mix_internal(AudioFrame *p_buffer, int p_fr } } +#ifdef DEBUG_ENABLED + if (input_ofs > input_position && (input_ofs - input_position) < (p_frames * 2)) { + print_verbose(String(get_class_name()) + " buffer underrun: input_position=" + itos(input_position) + " input_ofs=" + itos(input_ofs) + " input_size=" + itos(input_size)); + } +#endif + AudioDriver::get_singleton()->unlock(); } diff --git a/servers/audio/audio_stream.h b/servers/audio/audio_stream.h index 2740f86d550..f6ed45cc9ce 100644 --- a/servers/audio/audio_stream.h +++ b/servers/audio/audio_stream.h @@ -122,8 +122,6 @@ class AudioStreamPlaybackMicrophone : public AudioStreamPlaybackResampled { GDCLASS(AudioStreamPlaybackMicrophone, AudioStreamPlayback) friend class AudioStreamMicrophone; - static const int MICROPHONE_PLAYBACK_DELAY = 256; - bool active; unsigned int input_ofs; diff --git a/servers/audio_server.cpp b/servers/audio_server.cpp index d36057b4659..37454d92537 100644 --- a/servers/audio_server.cpp +++ b/servers/audio_server.cpp @@ -80,6 +80,14 @@ double AudioDriver::get_mix_time() const { return total; } +void AudioDriver::input_buffer_init(int driver_buffer_frames) { + + const int input_buffer_channels = 2; + input_buffer.resize(driver_buffer_frames * input_buffer_channels * 4); + input_position = 0; + input_size = 0; +} + void AudioDriver::input_buffer_write(int32_t sample) { input_buffer.write[input_position++] = sample; diff --git a/servers/audio_server.h b/servers/audio_server.h index ba6569eb389..b12ca6e5892 100644 --- a/servers/audio_server.h +++ b/servers/audio_server.h @@ -59,6 +59,7 @@ protected: void audio_server_process(int p_frames, int32_t *p_buffer, bool p_update_mix_time = true); void update_mix_time(int p_frames); + void input_buffer_init(int driver_buffer_frames); void input_buffer_write(int32_t sample); #ifdef DEBUG_ENABLED