KIOSAudioQualityResult Class Reference

#include <KIOSAudioQualityResult.h>

Instance Methods
(nonnull NSString *)	- toJson

Properties
NSArray< NSNumber * > *	frameRmsValues
NSUInteger	clippedSampleCount
NSNumber *	snrValue
NSNumber *	meanSpeechRmsValue
NSNumber *	meanNonSpeechRmsValue
NSNumber *	peakSpeechRmsValue
BOOL	initialSegmentRmsWarning

Detailed Description

AudioQualityResult class with specified rms values vector, clipped sample count and estimated SNR value.

AudioQualityResult will be computed during final response computation, and provided in an instance of KIOSResponse object.

Method Documentation

toJson

- (nonnull NSString *) toJson

Serialises contents of AudioQualityResult into JSON format.

Returns

A JSON string representation of AudioQualityResult.

Property Documentation

clippedSampleCount

- (NSUInteger) clippedSampleCount

readnonatomicassign

Number of raw samples in processed audio that were clipping, i.e. reaching either maximum or minimum value. Clipping indicate that the user might be too close to the microphone during audio capture or fidgeting with the microphone. It can have negative implications on speech recognition performance.

frameRmsValues

- (NSArray<NSNumber *>*) frameRmsValues

readnonatomicassign

Root mean square (RMS) values for each frame (25ms long with 10ms shift) in decibels in the processed audio. Values are computed as 20*log10(√(sum of squared sample values)). Where a frame has a zero signal level, the RMS dB value is reported as around -100dB.

initialSegmentRmsWarning

- (BOOL) initialSegmentRmsWarning

readnonatomicassign

Flag indicating that high RMS (root mean square) value was detected during the initial part of the processed audio. This could either indicate that: a) device is playing audio, which is then captured by the microphone); b) the user has already started speaking, and recognizer started to listen too late. c) high levels of noise in general.

meanNonSpeechRmsValue

- (NSNumber*) meanNonSpeechRmsValue

readnonatomicassign

Mean frame root mean square (RMS) level for the noise segments in processed audio, in decibels.

meanSpeechRmsValue

- (NSNumber*) meanSpeechRmsValue

readnonatomicassign

Mean frame root mean square (RMS) level for the speech segments in processed audio in decibels.

NOTE: This value will be nil if no speech segments were available in the response.

peakSpeechRmsValue

- (NSNumber*) peakSpeechRmsValue

readnonatomicassign

Estimated peak root mean square (RMS) level of speech in the processed audio, in decibels. This value is computed as 98th percentile of all the RMS speech levels to filter outliers. Low values would indicate faint speech or user being too far from the microphone. This metric may not work well for responses with very short speech segments.

NOTE: This value will be nil if no speech segments were available in the response.

snrValue

- (NSNumber*) snrValue

readnonatomicassign

Estimated signal to noise ratio (SNR) in decibels for processed audio. SNR is computed as a difference between meanSpeechRmsValue and mean_noise_rms_value. The way SNR is currently computed may not take transient noise as effectively into account as it does stationary background noise. Low values will affect speech recognition performance.

NOTE: snrValue will be nil if there wasn't sufficient data to compute it.

---

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

• KeenASR/KIOSAudioQualityResult.h