Skip to content

Audio: MFCC: Add Voice Activity Detection based on Mel spectrum #10782

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
singalsu wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

Audio: MFCC: Add Voice Activity Detection based on Mel spectrum #10782

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions src/arch/host/configs/library_defconfig
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,7 @@ CONFIG_COMP_IIR=y
CONFIG_COMP_IGO_NR=y
CONFIG_COMP_LEVEL_MULTIPLIER=y
CONFIG_COMP_MFCC=y
CONFIG_COMP_MFCC_VAD=y
CONFIG_COMP_MODULE_ADAPTER=y
CONFIG_COMP_MULTIBAND_DRC=y
CONFIG_COMP_MUX=y
Expand Down
3 changes: 3 additions & 0 deletions src/audio/mfcc/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,4 +5,7 @@ if(CONFIG_COMP_MFCC STREQUAL "m" AND DEFINED CONFIG_LLEXT)
add_dependencies(app mfcc)
else()
add_local_sources(sof mfcc.c mfcc_setup.c mfcc_common.c mfcc_generic.c mfcc_hifi4.c mfcc_hifi3.c)
if(CONFIG_COMP_MFCC_VAD)
add_local_sources(sof mfcc_vad.c)
endif()
endif()
11 changes: 11 additions & 0 deletions src/audio/mfcc/Kconfig
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,3 +24,14 @@ config COMP_MFCC
The characteristic of the audio features are defined in the binary
control blob. Directory tools/tune/mfcc contains a tool to create
the configurations.

config COMP_MFCC_VAD
bool "MFCC Voice Activity Detection"
depends on COMP_MFCC
default n
help
This option enables a Voice Activity Detector (VAD) that operates
on the Mel spectrum values produced by the MFCC component. The VAD
flag is inserted into the output stream as the first int32_t value
after the magic header word. The VAD tracks a per-bin noise floor
and detects speech using a weighted energy delta with hangover.
44 changes: 44 additions & 0 deletions src/audio/mfcc/mfcc_common.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,10 @@
#include <stddef.h>
#include <stdint.h>

#ifdef CONFIG_COMP_MFCC_VAD
#include <sof/audio/mfcc/mfcc_vad.h>
#endif

LOG_MODULE_REGISTER(mfcc_common, CONFIG_SOF_LOG_LEVEL);

/*
Expand Down Expand Up @@ -144,6 +148,10 @@ static int mfcc_stft_process(const struct comp_dev *dev, struct mfcc_comp_data *
sat_int32(Q_MULTSR_32X32(s, config->mel_scale, 23, 12, 23));
}

#ifdef CONFIG_COMP_MFCC_VAD
/* Run VAD on the mel log spectrum before further processing */
state->vad_flag = mfcc_vad_update(&cd->vad, state->mel_log_32);
#endif
/* Store Q9.7 version in mel_spectra for s16 output mode */
for (j = 0; j < state->dct.num_in; j++)
state->mel_spectra->data[j] =
Expand Down Expand Up @@ -289,6 +297,9 @@ void mfcc_s16_default(struct processing_module *mod, struct input_stream_buffer

state->out_remain = num_ceps;
state->magic_pending = true;
#ifdef CONFIG_COMP_MFCC_VAD
state->vad_pending = true;
#endif
}

/* Write to sink, limited by period size */
Expand All @@ -301,6 +312,15 @@ void mfcc_s16_default(struct processing_module *mod, struct input_stream_buffer
state->magic_pending = false;
}

#ifdef CONFIG_COMP_MFCC_VAD
/* Write VAD flag as first value after magic (as two int16_t = one int32_t) */
if (state->vad_pending && sink_samples >= 2) {
w_ptr = mfcc_sink_copy_data_s16(sink, w_ptr, 2, (int16_t *)&state->vad_flag);
sink_samples -= 2;
state->vad_pending = false;
}
#endif

/* Write cepstral/mel data from scratch buffer */
to_copy = MIN(state->out_remain, sink_samples);
if (to_copy > 0) {
Expand Down Expand Up @@ -392,6 +412,9 @@ void mfcc_s24_default(struct processing_module *mod, struct input_stream_buffer

state->out_remain = num_ceps;
state->magic_pending = true;
#ifdef CONFIG_COMP_MFCC_VAD
state->vad_pending = true;
#endif
}

/* Write to sink, limited by period size */
Expand All @@ -404,6 +427,15 @@ void mfcc_s24_default(struct processing_module *mod, struct input_stream_buffer
state->magic_pending = false;
}

#ifdef CONFIG_COMP_MFCC_VAD
/* Write VAD flag as first value after magic */
if (state->vad_pending && sink_samples >= 1) {
w_ptr = mfcc_sink_copy_data_s32(sink, w_ptr, 1, &state->vad_flag);
sink_samples -= 1;
state->vad_pending = false;
}
#endif

if (state->mel_only) {
/* Write 32-bit mel data Q9.15, one value per int32_t */
to_copy = MIN(state->out_remain, sink_samples);
Expand Down Expand Up @@ -467,6 +499,9 @@ void mfcc_s32_default(struct processing_module *mod, struct input_stream_buffer

state->out_remain = num_ceps;
state->magic_pending = true;
#ifdef CONFIG_COMP_MFCC_VAD
state->vad_pending = true;
#endif
}

/* Write to sink, limited by period size */
Expand All @@ -479,6 +514,15 @@ void mfcc_s32_default(struct processing_module *mod, struct input_stream_buffer
state->magic_pending = false;
}

#ifdef CONFIG_COMP_MFCC_VAD
/* Write VAD flag as first value after magic */
if (state->vad_pending && sink_samples >= 1) {
w_ptr = mfcc_sink_copy_data_s32(sink, w_ptr, 1, &state->vad_flag);
sink_samples -= 1;
state->vad_pending = false;
}
#endif

if (state->mel_only) {
/* Write 32-bit mel data Q9.23, one value per int32_t */
to_copy = MIN(state->out_remain, sink_samples);
Expand Down
18 changes: 18 additions & 0 deletions src/audio/mfcc/mfcc_setup.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,10 @@
#include <stddef.h>
#include <stdint.h>

#ifdef CONFIG_COMP_MFCC_VAD
#include <sof/audio/mfcc/mfcc_vad.h>
#endif

/* Definitions for cepstral lifter */
#define PI_Q23 Q_CONVERT_FLOAT(3.1415926536, 23)
#define TWO_PI_Q23 Q_CONVERT_FLOAT(6.2831853072, 23)
Expand Down Expand Up @@ -346,10 +350,24 @@ int mfcc_setup(struct processing_module *mod, int max_frames, int sample_rate, i
state->waiting_fill = true;
state->prev_samples_valid = false;
state->magic_pending = false;
#ifdef CONFIG_COMP_MFCC_VAD
state->vad_pending = false;
state->vad_flag = 0;
#endif
state->out_data_ptr = NULL;
state->out_data_ptr_32 = NULL;
state->out_remain = 0;

#ifdef CONFIG_COMP_MFCC_VAD
ret = mfcc_vad_init(&cd->vad, config->num_mel_bins, sample_rate);
if (ret < 0) {
comp_err(dev, "Failed VAD init");
goto free_lifter;
}

comp_info(dev, "VAD enabled, num_mel_bins = %d", config->num_mel_bins);
#endif

comp_dbg(dev, "done");
return 0;

Expand Down
Loading
Loading