Ch17 - sound quality

where does sound quality matter?

Audio / HiFi

Concert halls

Noise annoyance

Speech intelligibility / Public address

Hearing aids / cochlear implants

Razors / Car engine sounds / Mechanical products

Musical instruments

sound quality is also important when making different technical choices (what effect will they have on quality)

quantity vs quality?

quantity - categorisedd by type or class of object where 2 ovbservations or entities cant be compared on same metric scale (1m vs 90.5kg)

quality - grade or rank objects on a subjective scale of preferability such as good – poor

how does sound quality relate to psychoacoustics?

Classical psychoacoustics sees human as simple metering device though basic results are of course very useful

In sound quality the expectations, mood, and other cognitive factors are important

Judged sound quality for devices or communication channels depends heavily on expectations

Basic psychoacoustic experimentation techniques may not be used

how is sound quality measured and evaluated?

sound quality → subjective but can be approximated by objective & computational criteria

Subjective quality → evaluated by listening experiments

• compare to "perfect quality" reference to find out if any degradation can be noticed

• compare two or more sounds and sort then by quality preference

• give an overall quality rating on a numerical scale

Based on subjective experimentation, a computational (objective) measure and model → derived to simulate the perceived quality

• Objective measures are less laborious and yield high repeatability

• its important to check the validity range of a model

systemic framework for sound quality?

how to rate subjective sound quality

Mean opinion score (MOS)

Principle: ask opinion of some aspect of sound quality in numerical scale with anchors, take an average + other statistical measures

MOS scales have been also defined for measurement of improvement or degradation of quality

Multiple-stimulus hidden reference with anchors (MUSHRA) (group project core)

what is monaural audio quality?

Degradations of audio present by listening only one channel

(imagine audio engineer has mastered audio content as intended, and degradations in transmission channel are measured, not diffs in perceptions)

Deviations in:

• Magnitude response

• Phase and group delays

• Non-linear distortions

• Signal-to-noise ratio

how can we estimate perceived difference?

auditory models (wow)

what is Perceptual audio quality measure (PAQM)?

objective metric for assessing audio quality that uses a model of the human auditory system to predict how a person would perceive the sound

what is PEAQ (Perceptual audio quality)?

standardized algorithm for objectively measuring perceived audio quality (improved PAQM)

spatial audio quality?

accounts for 30% of total quality

loses quality if sound degraded by colourations

how to measure subjectively?

• compare with reference, if exists and can be brought to listening room (no simple solution exists tho)

how to measure objectively?

• binaural auditory models under research, tho standardisation has failed

what determines quality of speech communication?

Speech intelligibility

Speaker recognizability

Speech naturalness

Subjective and objective measurement

what determines quality of speech communication (subjectively)?

• Articulation tests and articulation score (/CV/ or /CVC/ sequences used to measure recognition percentage)

• Intelligibility test and intelligibility score (recognition percentage using meaningful words or sentences)

• Rhyme tests (RT)

• Diagnostic rhyme tests (DRT) (modifying single distinctive feature at a time (nasality, voicing, etc.) in RT)

• Speech interference tests (find a disturbing noise level of 50% articulation)

• Quality comparison method, including pairwise comparison methods

Other methods

• Communicability tests (communicate a drawing task, measure the difficulty)

• Task recall tests (memorizing ability)

• Noise suppression test

what determines quality of speech communication (objectively)?

• Articulation index (AI) (for measuring a (linear) speech transmission channel with additive noise articulation loss is assumed to be additive from 20 frequency bands)

• Percentage articulation loss of consonants (%ALcons) (measure of speech intelligibitly, estimated from room acoustic parameters)

• Room acoustical indices

• Speech transmission index (STI, STIPA) (based on modulation transfer function)

• Signal-to-noise ratio (SNR) (ratio of speech vs. noise (power) level (in dB))

• Spectral distance measures

• Auditory sound quality measures (based on auditory modeling)

what is modulation transfer function?

each critical band in auditory system is analysed by signal level (modulation envelope)
More important than the exact transfer function is modulation transfer function, i.e., how signal modulations in each critical band are transmitted

modulation transfer degraded by 

• reverberation (lowpass of modulation) → depth of modulation changes due to reverberation and noise

• background noise (reduction of relative modulation)

• these effects are multiplicative (cascaded)

Reverberation reduces the modulation depth more at high modulation frequencies than at low modulation frequencies. Noise, in turn, reduces the modulation depth equally at all modulation frequencies. The rightmost plots show the corresponding modulation transfer functions for cases A and B. In case A, the reverberation clearly acts as a low-pass filter in the transfer function, while the noise in case B reduces the modulation evenly at all modulation frequencies.

what is Speech Transmission Index STI

Measure m with many carrier frequencies and modulation frequencies, take logarithm-like-measure and take a weighted average

optimally reflects the intelligibility of speech over measured channel

STI measurement requires presentation of all carrier-frequency-pairs at separate times – slow

measure of speech transmission quality

STI measures some physical characteristics of transmission channel (a room, electro-acoustic equipment, telephone line, etc.), and expresses ability of channel to carry across characteristics of speech signal

Objective speech quality measurement for telecommunication?

STI doesnt estimate naturalness of speech

Mobile telecommunication codecs may make speaker unidentifiable, although intelligibility is good

more methods needed

• Models for general speech quality are expected to give a high MOS value only for natural-sounding and intelligible speech

• Methods for measuring the perceptual effect of background noise suppression

• Measures for echo suppression

if a mobile device enters market, sound quality should be tested but listenin tests would be tedious (Basic quality, Effects of network problems delays, echos, background noise, background noise suppression), so standerdised auditory model based evaluation tool is used instead (3GPP - global standards organization that develops technical specifications for mobile communication, including detailed acoustic testing standards for mobile devices)

Techniques for objective speech quality evaluation?

• Perceptual speech quality measure (PSQM)

• Perceptual evaluation of speech quality (PESQ)

• Telecommunications objective speech quality assessment (TOSQA)

• Perceptual objective listening quality assessment) POLQA

• Hearing-aid speech quality index (HASQI)

devices entering markets need to produce a high enough score with one of those metrics

Techniques for evaluation of the effect of background noise?

mobile phones often involve algorithms to suppress background noise using non-linear DSP methods with time-variant processing

• Non-linear noise-suppression algorithms try to reduce noise

• How much does the algorithm reduce the problem

Standardized MOS scales for listening tests

Similar techniques exist also for measurement of the effect of echoes in two-way communication channel

what are some perceptual attributes of concert halls?

Defined by Beranek by personal listening,

• Intimacy or presence / Reverberation or liveness / Spaciousness: Apparent source width(ASW) / Spaciousness: Listener envelopment (LEV) / Clarity / Warmth / Loudness / Acoustic glare / Brilliance / Balance / Blend / Ensemble / Immediacy of response / Texture / Freedom from echo / Dynamic range and background noise level / Extraneous effects on tonal quality / Uniformity of sound

List is subject to debate and further studies

Objective measures of concert hall acoustics?

Measure few impulse responses in the hall

Compute the values from the responses

Subject to criticism, correspondence to actual perceptual quality is questionable

Noise quality?

Sound that is disturbing or annoying → purely subjective measure

Annoyance → depends on signal and on context (speech and laughter are disturbing in open-plan office, ventilation humming is ok)

Disturbance (Speech intelligibility over distance is not desired in open-plan office opposite to theatres and auditoria)

product sound quality?

Minimize negative effects and maximize positive effects of product sound

Mechanic devices communicate their functioning to to the user

Electronic devices may have loudspeakers to do the same (car turn

signal earlier relay, nowadays loudspeaker)

Examples:

• Cars and work machines

• Home appliances

• Office equipment

• Personal devices