voicesdk::diar::DiarizationEngine Class Reference

Diarization engine class (interface), intended to perform speaker diarization. More...

#include <voicesdk/diarization/diarization.h>

Public Types
using	Ptr = std::shared_ptr< DiarizationEngine >

Public Member Functions
virtual TimeStamps	GetSegmentation (const float *float_samples, const size_t samples_num, const size_t sample_rate, const size_t num_speakers=0)=0
	Performs speaker diarization from the given float (in [-1; 1] range) audio samples. More...
virtual TimeStamps	GetSegmentation (const int16_t *pcm16_samples, const size_t samples_num, const size_t sample_rate, const size_t num_speakers=0)=0
	Performs speaker diarization from the given PCM16 audio samples. More...
virtual TimeStamps	GetSegmentation (const uint8_t *pcm16_bytes, const size_t bytes_num, const size_t sample_rate, const size_t num_speakers=0)=0
	Performs speaker diarization using given PCM16 samples bytes representation. More...
virtual TimeStamps	GetSegmentation (const std::string &audio_path, const size_t num_speakers=0)=0
	Performs speaker diarization from the given audio file. More...

Static Public Member Functions
static DiarizationEngine::Ptr	Create (const std::string &init_dir)
	Creates DiarizationEngine instance. More...

Detailed Description

Diarization engine class (interface), intended to perform speaker diarization.

Member Function Documentation

Create()

static DiarizationEngine::Ptr voicesdk::diar::DiarizationEngine::Create ( const std::string &  init_dir )

static

Creates DiarizationEngine instance.

Parameters

init_dir

path to initialization data directory

Returns

Smart pointer to created DiarizationEngine instance

Exceptions

std::runtime_error	if runtime error occurred
voicesdk::LicenseException	if license error occurred

GetSegmentation() [1/4]

virtual TimeStamps voicesdk::diar::DiarizationEngine::GetSegmentation ( const float *  float_samples, const size_t  samples_num, const size_t  sample_rate, const size_t  num_speakers = 0  )

pure virtual

Performs speaker diarization from the given float (in [-1; 1] range) audio samples.

Parameters

float_samples	pointer to array with samples
samples_num	size of array with samples
sample_rate	sample rate
num_speakers	optional parameter, can be passed if number of speakers in record is known (e.g. 2 for phone conversation), increases diarization accuracy; if 0, then engine tries to determine number of speakers by itself, thus accuracy can be lower

Exceptions

std::runtime_error	if runtime error occurred
voicesdk::LicenseException	if license error occurred

Returns

timestamps of speakers utterances

GetSegmentation() [2/4]

virtual TimeStamps voicesdk::diar::DiarizationEngine::GetSegmentation ( const int16_t *  pcm16_samples, const size_t  samples_num, const size_t  sample_rate, const size_t  num_speakers = 0  )

pure virtual

Performs speaker diarization from the given PCM16 audio samples.

Parameters

pcm16_samples	pointer to array with samples
samples_num	size of array with samples
sample_rate	sample rate
num_speakers	optional parameter, can be passed if number of speakers in record is known (e.g. 2 for phone conversation), increases diarization accuracy; if 0, then engine tries to determine number of speakers by itself, thus accuracy can be lower

Exceptions

std::runtime_error	if runtime error occurred
voicesdk::LicenseException	if license error occurred

Returns

timestamps of speakers utterances

GetSegmentation() [3/4]

virtual TimeStamps voicesdk::diar::DiarizationEngine::GetSegmentation ( const std::string &  audio_path, const size_t  num_speakers = 0  )

pure virtual

Performs speaker diarization from the given audio file.

Parameters

audio_path	path to audio file
num_speakers	optional parameter, can be passed if number of speakers in record is known (e.g. 2 for phone conversation), increases diarization accuracy; if 0, then engine tries to determine number of speakers by itself, thus accuracy can be lower

Exceptions

std::runtime_error	if runtime error occurred
voicesdk::LicenseException	if license error occurred

Returns

timestamps of speakers utterances

GetSegmentation() [4/4]

virtual TimeStamps voicesdk::diar::DiarizationEngine::GetSegmentation ( const uint8_t *  pcm16_bytes, const size_t  bytes_num, const size_t  sample_rate, const size_t  num_speakers = 0  )

pure virtual

Performs speaker diarization using given PCM16 samples bytes representation.

Parameters

pcm16_bytes	pointer to array with bytes
bytes_num	size of array with bytes
sample_rate	sample rate
num_speakers	optional parameter, can be passed if number of speakers in record is known (e.g. 2 for phone conversation), increases diarization accuracy; if 0, then engine tries to determine number of speakers by itself, thus accuracy can be lower

Exceptions

std::runtime_error	if runtime error occurred
voicesdk::LicenseException	if license error occurred

Returns

timestamps of speakers utterances