1 Audio Coding

2 Digitization Processing Signal encoder Signal decoder samplingquantization storage Analog signal Digital data

3 Overview of Today PCM –Linear –  -LaW DPCM ADPCM MPEG-1 Vocoding Sampling Techniques Generic Coding Techniques Psychoacoutic Coding Speech Specific Techniques

4 Encode Design Bandlimiting filter –Smooth analog signals Analog to digital converter (ADC) –Sample and Quantize analog signals.

5 Bandlimiting filter Pass only frequency components up to half of Nyquist rate.

6 Analog to digital converter

7 Sampling Pulse Amplitude Modulation (PAM) –Each sample ’ s amplitude is represented by 1 ________ value Sampling theory (_________) –If input signal has ________ frequency (bandwidth) f, sampling frequency must be at least ____ –With a _____-pass filter to interpolate between samples, the input signal can be fully reconstructed

8 PCM Pulse Code Modulation (PCM) –Each sample ’ s amplitude represented by an ________ code-word –Each bit of resolution adds __ dB of dynamic range –Number of bits required depends on the amount of noise that is tolerated Quantization error (“noise”) n = SNR –

9 Linear PCM Quantization levels are _________ spaced. ___ bit samples provide plenty of dynamic range. Compact Disks do this.

10 Under Sampling Sample rate under Nyquist rate LF also called antialiasing filter Added to original signal and cause distortion.

11 Quantization intervals

12 Associated waveform set

13   -Law companding (ITU Rec. G.711) Non-linear quantization of the signal ’ s amplitude –Quantization step-size decreases logarithmically with signal ______ –Low-amplitude samples represented with ______ accuracy than high-amplitude samples –Humans are less sensitive to changes in “ ____ ” sounds than “ _____ ” sounds

14 f(x) = 127 x sign(x) x ln(1 +  |x|) ln(1 +  ) (x normalized to [-1, 1])   -Law companding Provides __-bit quality (dynamic range) with an _-bit encoding Used in North American & Japanese ISDN voice service Simple to compute encoding

Difference Encoding Differential-PCM (DPCM) –Exploit _________ redundancy in samples –___________ between 2 x-bit samples can be represented with significantly fewer than x-bits –Transmit the difference (rather than the ________)

16 DPCM Working Principle Previous sampling value

“Slope Overload” Slope Overload Problem Differences in high frequency signals near the ___________ frequency cannot be represented with a smaller number of bits! –Error introduced leads to severe distortion in the ______ frequencies

18 Adaptive DPCM (ADPCM) Use a larger step-size to encode differences between ______-frequency samples & a smaller step-size for differences between ____- frequency samples Use ________ sample values to estimate changes in the signal in the near future

Predictor + – + y-bit PCM sample x-bit ADPCM “difference” Difference Quantizer Step-Size Adjuster Dequantizer + Predicted PCM Sample n+1 ADPCM To ensure differences are always small... –Adaptively change the ____-size (quanta) –(Adaptively) attempt to _____ next sample value

20 Psychoacoustic Fundamentals Absolute threshold of hearing Critical band frequency analysis Frequency masking Temporal masking

21 Absolute Threshold of Hearing Human perception of sound is a function of ________ and signal __________ –(MPEG exploits this relationship.) Sampled segments of the source audio waveform are analyzed but only those features _____________ to the ear are transmitted. Psychoacoustic model is used to identify _________ masking and ________ masking and eliminate them from the transmitted signal. Sound Level (dB) Frequency (kHz) Inaudible Audible Maximum allowable Energy level for Coding distortion

Sound Level (dB) Frequency (kHz) Inaudible Audible Masking tone Masked tone Auditory Masking The presence of tones at certain frequencies makes us unable to perceive tones at other “ _________ ” frequencies –Humans cannot distinguish between tones within _____ Hz at low frequencies and _____kHz at high frequencies

23 MPEG Encoder Block Diagram MappingQuantizerCoding Frame Packing Psycho- acoutstic Model PCM Audio Samples (32, 44.1, 48 kHz) Encoded Bitstream Ancillary Data

24 Vo-coding Concept: Develop a __________ model of the vocal cords & throat –Derive/compute _____ parameters for a short interval and transmit to the decoder –Use the parameters to _______ speech at the decoder So what is a good model? –A “ buzzer ” in a “ tube ” ! –The buzzer is characterized by its _________ & _______ –The tube is characterized by its ___________s

Amplitude Frequency (kHz) Vocoding - Basic Concepts Formant — frequency maxima & minima in the spectrum of the speech signal Vocoders code –_____ –Period –_________, and –signaling vocal tract _________ parameters Voiced sounds, m,v,and l. Unvoiced sounds, f and s.

26 “yadda yadda yadda” y(n) = a k y(n – k) + G x x(n)  k=1 p Linear Predictive Coding (LPC) –A sample is represented as a linear combination of ___ previous ________s “ Buzzer ” and “ Tube ” Model Vocoding principles: –voice = _________s + buzz ______ & intensity –voice – estimated ________s = “ residue ”

27 LPC Decoder artificially generates speech via _________ synthesis –A mathematical simulation of the _______ as a series of bandpass filters –Encoder codes & transmit filter _______, pitch period, gain factor, & nature of excitation

28 LPC Schematic

29 LPC Related Standards Standards: –Regular Pulse Excited Linear Predictive Coder (RPE-LPC) Digital cellular standard GSM 6.1 (___ kbps) –Code Excited Linear Predictive Coder (CELP) US Federal Standard 1016 (_____ kbps) Waveform template based to improve sound quality. –Linear Predictive Coder (LPC) US Federal Standard 1015 (______ kbps) Very synthetic and used primarily in military applications with very limited bandwidth.

30 Networking Concerns Audio bandwidth is actually quite small. But human sensitivity to loss and noise is quite ________. Networking concerns: –_______ concealment –________ control Especially for telephony applications.