TfLiteMicroSpeechRecognizeCommands Class Reference

This class is designed to apply a very primitive decoding model on top of the instantaneous results from running an audio recognition model on a single window of samples. It applies smoothing over time so that noisy individual label scores are averaged, increasing the confidence that apparent matches are real. To use it, you should create a class object with the configuration you want, and then feed results from running a TensorFlow model into the processing method. The timestamp for each subsequent call should be increasing from the previous, since the class is designed to process a stream of data over time. More...

#include <TfLiteAudioStream.h>

Inheritance diagram for TfLiteMicroSpeechRecognizeCommands:

Classes
struct	Result

Public Member Functions
	TfLiteMicroSpeechRecognizeCommands ()
bool	begin (TfLiteConfig cfg) override
	Setup parameters from config.
virtual TfLiteStatus	getCommand (const TfLiteTensor *latest_results, const int32_t current_time_ms, const char **found_command, uint8_t *score, bool *is_new_command) override

Protected Member Functions
int	categoryCount ()
	Determines the number of categories.
void	deleteOldRecords (int32_t limit)
	Removes obsolete records from the queue.
TfLiteStatus	evaluate (const char **found_command, uint8_t *result_score, bool *is_new_command)
	Finds the result.
int	resultCategoryIdx (int8_t *score)
	finds the category with the biggest score
TfLiteStatus	validate (const TfLiteTensor *latest_results)
	Checks the input data.

Protected Attributes
TfLiteConfig	cfg
int32_t	current_time_ms =0
int	previous_cateogory =-1
int32_t	previous_time_ms =0
Vector< Result >	result_queue
int32_t	time_since_last_top =0

Detailed Description

This class is designed to apply a very primitive decoding model on top of the instantaneous results from running an audio recognition model on a single window of samples. It applies smoothing over time so that noisy individual label scores are averaged, increasing the confidence that apparent matches are real. To use it, you should create a class object with the configuration you want, and then feed results from running a TensorFlow model into the processing method. The timestamp for each subsequent call should be increasing from the previous, since the class is designed to process a stream of data over time.

Constructor & Destructor Documentation

TfLiteMicroSpeechRecognizeCommands()

TfLiteMicroSpeechRecognizeCommands ( )

inline

Member Function Documentation

begin()

bool begin ( TfLiteConfig  cfg )

inlineoverridevirtual

Setup parameters from config.

Implements TfLiteAbstractRecognizeCommands.

categoryCount()

int categoryCount ( )

inlineprotected

Determines the number of categories.

deleteOldRecords()

void deleteOldRecords ( int32_t  limit )

inlineprotected

Removes obsolete records from the queue.

evaluate()

TfLiteStatus evaluate ( const char **  found_command, uint8_t *  result_score, bool *  is_new_command  )

inlineprotected

Finds the result.

getCommand()

virtual TfLiteStatus getCommand ( const TfLiteTensor *  latest_results, const int32_t  current_time_ms, const char **  found_command, uint8_t *  score, bool *  is_new_command  )

inlineoverridevirtual

Implements TfLiteAbstractRecognizeCommands.

resultCategoryIdx()

int resultCategoryIdx ( int8_t *  score )

inlineprotected

finds the category with the biggest score

validate()

TfLiteStatus validate ( const TfLiteTensor *  latest_results )

inlineprotected

Checks the input data.

Member Data Documentation

cfg

TfLiteConfig cfg

protected

current_time_ms

int32_t current_time_ms =0

protected

previous_cateogory

int previous_cateogory =-1

protected

previous_time_ms

int32_t previous_time_ms =0

protected

result_queue

Vector<Result> result_queue

protected

time_since_last_top

int32_t time_since_last_top =0

protected

---

The documentation for this class was generated from the following file:

• src/AudioTools/AudioLibs/TfLiteAudioStream.h