Instructor: A. Spanias
We use the Book Audio Signal Processing and Coding (see below) (Amazon)
This two-day course describes the fundamental principles, techniques, and algorithms used in current applications; including a detailed discussion of current speech coding and telecommunication standards. The course starts with a discussion on: the basic speech representation methods, the performance measures used to evaluate coded speech, and the role of the standards. Algorithm descriptions include: ADPCM, sub-band coding, adaptive transform coding, sinusoidal transform coding (STC), multiband excitation (MBE) coding, linear predictive coding (LPC), residual excited LPC, and analysis-by-synthesis LPC (multipulse LPC, regular pulse LPC, code excited LPC). The course includes lecture, audio, and computer demonstrations of recent speech coding standards including voice-over IP algorithms. The course also describes wideband audio standards such as the AC-3, the MPEG (MP3) audio layers, and the Sony systems algorithms. Participants will get a copy of all the viewgraphs as well as a package of detailed notes for the course. Bonus: Participants will also get an opportunity to experiment with software for the FS1015 LPC-10 and the FS1016 CELP algorithms. Participants take back a copy of this software to the work place.
The course is designed for engineers and managers who need to understand emerging speech coding techniques and telecommunication standards. The course should be of particular interest to engineers of telecommunication and computer companies who are evaluating new digital systems for mobile telephones, multimedia computing, and wireless teleconferencing. Participants should have an understanding of basic engineering mathematics. Individuals may want to take the DSP course to obtain a broader perspective on signal and speech processing.
General: characterizations of voiced and unvoiced speech, analysis/synthesis models, performance measures and complexity issues - DSP Background: digital filters for speech processing, random signals, autocorrelation and covariance methods, FFT processing of speech - Waveform Coders: scalar and vector quantization, the ADPCM ITU G.726, sub-band and transform Coders, the ITU G.722 sub-band coder - Speech Coding Using Sinusoidal Analysis/Synthesis Models: the sinusoidal transform coder (STC), the multiband excitation coder (MBE), The Inmarsat Standard, IRIDIUM IBME/AMBE - Vocoder Methods: the channel, formant, and homomorphic vocoders, linear predictive vocoders, the LPC-10 Algorithm, mixed excitation LPC, RELP - Analysis-by-Synthesis Linear Predictive Coders: multi-pulse excited linear prediction, the Skyphone standard, Regular Pulse Excitation (RPE) coders, the Full-Rate, half-rate, and Enhanced full-rate GSM standards, code excited linear prediction, Algebraic CELP, Relaxed CELP, the IS-54 and IS-641 (PCS) standards, the Federal Standard 1016 CELP, the ITU G.728 low-delay CELP coder, the ACELP ITU G.729 and G.729A,, the CDMA QCELP (IS-96), the CDMA EVRC (IS-127), True Speech G723.1, G.4, CDMA 2000 - SMV, Adaptive Multirate Coder for GSM and the new Federal Standard 1017 2.4 kbit/s Mixed Excitation Coder. Info also on proposed vocoders for CDMA 2000, ITU-4, and Adaptive Multi-rate GSM standard. Audio coding topics include introduction to perceptual coding, the MPEGs (MP3), SDDS, DTS, AC-3, and other algorithms embedded in recent streaming audio, cinema, and digital audio products.
Participants will learn about:
If you need information on any of the above (date/location/in house) send email at spanias@asu.edu.