(FM) Synthesis

From Synthesis to Sampling

Evolution of PC Audio

Proskauer—Yakovlev

Proskauer - Yakovlev

#1 Lec # 66 Fall 2017 12-14-2017 Overview

• Role of an audio interface • History of audio interfaces – PC Speaker (1981) to Intel HD Audio (2004) – Shift from synthesized audio to sampled audio • Modern audio devices – Entirely sample-based – Move toward integrated onboard audio – External DACs and amps

Proskauer - Yakovlev

#2 Lec # 66 Fall 2017 12-14-2017 Role of an Audio Interface

• Perform audio recording and playback processing • Move the load of audio processing away from the CPU • Signal conversion – DAC for playback – ADC for recording

Proskauer - Yakovlev

#3 Lec # 66 Fall 2017 12-14-2017 Synthesis

• Synthesis is the process of electrically creating sound • There are many modes of synthesis • Frequency Modulation (FM) synthesis – The frequency of a carrier wave is modulated by another oscillator – Complex and unique waveforms can be produced – Particularly good at metallic and percussive sounds • Sample-based synthesis – Similar to synthesis using the standard waveforms but the seed waves are sampled sounds or instruments

Proskauer - Yakovlev

#4 Lec # 66 Fall 2017 12-14-2017 Sampling and Pulse Code Modulation

• Analog signals are continuous and have a specific value at every given moment in time • This type of data is not conducive to the way computers operate • Sampling is the process of approximating an analog signal by reading the value at a given interval • The quality of the sample is determined by the sample rate and the bit depth (resolution) • Pulse Code Modulation (PCM) is a method to digitally represent sampled audio

Proskauer - Yakovlev

#5 Lec # 66 Fall 2017 12-14-2017 PC Speaker (1981)

• First appeared in the IBM 5051 as a means of indicating motherboard error codes • Produced only a single square wave at a given frequency • Entirely CPU controlled, required lots of CPU performance to operate • Rapidly switching between frequencies could give the illusion of multiple voices

Proskauer - Yakovlev

#6 Lec # 66 Fall 2017 12-14-2017 PC Speaker PWM

• Pulse-width Modulation could be used to reproduce rudimentary sampled audio • The Programmable Interval Timer (PIT) would be configured to produce a PWM signal at a frequency higher than the speaker could reproduce • Adjusting the pulse width created a DC offset, extruding the speaker a specific amount • 6 bits of freedom • Very CPU intensive, required nearly all the CPU time – Graphics would lock up while audio was playing

Proskauer - Yakovlev

#7 Lec # 66 Fall 2017 12-14-2017 PCjr/Tandy 1000 (1983)

• PCjr was IBM’s first step into the home micro market • Included a Texas Instruments SN76496 programmable sound generator (PSG) – Capable of producing 3 square wave channels and one noise channel – 16 volume levels for each channel – Similar to the MOS Technologies SID chips on the Commodore 64

Proskauer - Yakovlev

#8 Lec # 66 Fall 2017 12-14-2017 AdLib (1987)

• First major improvement in PC audio • Featured the Yamaha YM3812 (OPL2) FM chip – 6 melodic voices with 5 percussion voices, or 9 melodic voices – 2 FM operators – Full ADSR envelope for each voice • Very high support in games from the time

Proskauer - Yakovlev

#9 Lec # 66 Fall 2017 12-14-2017 Creative Music System / Game Blaster (1988)

• First attempt from Creative Labs to produce a card • Featured two Philips SAA1099 PSGs – 12 channels of square waves • Marketed through RadioShack as the Game Blaster • Did not sell well since the AdLib card offered far better music quality

Proskauer - Yakovlev

#10 Lec # 66 Fall 2017 12-14-2017 Sound Blaster (1989)

• Designed to directly compete with the AdLib card • First card to feature sampling – 22.05kHz 8-bit mono playback – 13kHz 8-bit mono recording • Included the same Yamaha music chip as the AdLib • Commercial success – Integrated game port – Massive library of supported games – Pushed AdLib out of the market within 2 years

Proskauer - Yakovlev

#11 Lec # 66 Fall 2017 12-14-2017 Roland MT-32 (1987)

• External MIDI synthesizer module – 32 voices – 128 instruments – 30 percussion – 10 digital effects • Not originally intended for use in PC gaming – Very cost prohibitive, required MIDI interface card (typically the MPU-401) – Was created by Roland for amateur musicians as a budget external synthesizer • Best sounding audio solution at the time, supported by games of the time

Proskauer - Yakovlev

#12 Lec # 66 Fall 2017 12-14-2017 Gravis Ultrasound (1991)

• First PC card to support sample-based synthesis • Gravis GF1 – 256kB RAM – 32 voices – 16-bit playback – 8-bit record – 44.1kHz – Stereo • Supported hardware mixing, which was usually performed by the CPU • Did not contain FM synthesis chip, used the GF1 to emulate the OPL2

Proskauer - Yakovlev

#13 Lec # 66 Fall 2017 12-14-2017 AC’97 (1997)

• Audio codec developed by Intel in response to ubiquity of sound cards • Digital controller (DC97) – Built into southbridge of motherboard – Interfaced with CPU and system memory • AC-Link – Digital interface between DC97 and the codec • Codec – Third party chip that interfaced with the AC-Link and contained the DAC components

Proskauer - Yakovlev

#14 Lec # 66 Fall 2017 12-14-2017 Intel High Definition Audio (2004)

• Successor to AC’97 • Same controller/link/codec architecture as AC’97 • Specs – 15 input, 15 output streams – 16 PCM channels/stream – 8-32 bit sample resolution – 6-192kHz sample rate – Audio jack detection and retasking

Proskauer - Yakovlev

#15 Lec # 66 Fall 2017 12-14-2017 Sound Cards Today

• Onboard solutions have sufficient isolation to prevent EMI. – Placed far from other components • Discrete sound cards are less common • Discrete cards typically use PCIe ports – Offer more channels – Support for higher impedance headphones – Less EMI than onboard audio

Proskauer - Yakovlev

#16 Lec # 66 Fall 2017 12-14-2017 External DACs and Amplifiers

• Used typically in music production and professional audio work – Situated outside of a computer – Offer lowest possible EMI for clean sound – Amplifiers may be necessary for high impedance speakers – Minimizes THD

Proskauer - Yakovlev

#17 Lec # 66 Fall 2017 12-14-2017 Conclusion

• Early computers were not powerful enough to reproduce sampled audio – Synthesis was used to offload the audio generation workload off of the CPU

• Improvements in memory, processing power, and architecture bits allowed for dedicated sound cards to reproduce sampled audio

• Continued CPU performance improvements allowed audio generation to return to the CPU and motherboard- based chips

Proskauer - Yakovlev

#18 Lec # 66 Fall 2017 12-14-2017 References

Audio Codec ‘97, Intel Corporation, Revision 2.3, April, 2002.

Göhler, Stefan. “Phenomenal.” Crossfire Designs, Mar. 12, 2003, www.crossfire- designs.de/?lang=en&what=articles&name=showarticle.htm&article=soundcards.

Grosse, Robert. “Specifications of the IBM 5150 PC.” University of Saskatchewan Computer Museum, University of Saskatchewan, 22 Feb. 2012, computermuseum.usask.ca/articles-and-press/IBM- 5150-Specifications.pdf.

High Definition Audio Specification, Intel Corporation, Revision 1.0a, June 17, 2010.

Latimer, Joey. “Hit It, Maestro!” Compute, Apr. 1990, pp. 22–31.

Leonard, Jim. “IBM PC Ramblings.” That Oldskool Beat, Oldskool.org, 17 June 2016, www.oldskool.org/sound/pc.

Murray, David. “How Oldschool Sound/Music worked.” YouTube, YouTube, 5 Oct. 2015, www.youtube.com/watch?v=q_3d1x2VPxk.

“Electromagnetic Interference.” Wikipedia, Wikimedia Foundation, 6 Dec. 2017, en.wikipedia.org/wiki/Electromagnetic_interference.

Proskauer - Yakovlev

#19 Lec # 66 Fall 2017 12-14-2017