Over the years, the line between so-called consumer audio and professional audio gear has blurred as the number of features considered necessities for even basic audio products has grown to absorb most, or all, of the growing number of MIPS available from moderately-priced DSPs. Nevertheless, some general consensus has emerged on what some of the hottest markets for DSP-based consumer audio products will be for the next couple of years.
Home theater systems and audio/video receivers equipment are a traditional and very large market for DSPs where they play multiple roles in decoding, processing and delivering a first-class audio experience. First, they must decode the audio channels from the data stream of a DVD/Blu-ray player. The Blu-ray audio standard is especially demanding since it moves from the Dolby 5.1 channel encoding (and optional DTS encoding) that was used by DVDs to 7.1 channels encoded in several more complex standards, and at significantly higher data rates. Although there are several low-loss codecs that deliver excellent sound, the lossless standards from Dolby (Dolby TrueHD) and DTS (DTS Master Audio) are rapidly becoming the medium of choice for Blu-ray audio. Sampled at 192 kHz with 24-bit resolution, both deliver audio that’s bit-for-bit identical to the to the actual studio master. It’s no wonder that a fully-loaded decoder must be able to handle rates as high as 24.5 Mbit/s.
But that’s only half the story since DSPs also play a key role in automatic equalization, bass management, and so-called smart volume capabilities that keep the apparent volume of the amp at the same level – even when the input source changes. Although these features were originally developed for top-line pro-sumer gear, the declining cost of DSP power has made them nearly standard on mid-level and lower-priced A/V receivers. Basic automatic equalization capabilities are even starting to show up in the audio systems of some larger television receivers.
Some of these audio processing techniques are also being used to enhance the sound quality and spatial separation of smaller speakers found in media player docking stations, bookshelf audio systems, and the new sound bar systems that are beginning to challenge the traditional multi-cabinet home theater sound systems. Intended to be a simple, low-cost way to enhance the audio quality of flat panel TVs, a sound bar can vary from a simple two-speaker assemblies to multi-driver systems with as many as 16 speakers that produce home theater-like sound quality. Sound bars rely on DSPs to decode the Dolby 5.1/7.1 signals, virtualize the surround channels, and then downmix them to the available number of speakers. Further processing gives the closely-spaced speakers synthetic spatial separation, and can add niceties like bass and volume management.
Most major DSP manufacturers offer several hardware/software platforms that are tailored specifically to meet the performance requirements for the various price points along the consumer audio spectrum. Recent years have seen chip makers placing an emphasis on providing software tools, reference designs, and pre-packaged configurable IP modules that speed development time and make development of custom value-add functions easier.Meet the Players
Texas Instruments addresses most of its high- and mid-level consumer audio applications with its 32/64-bit, floating-point TMS320DA7xx Aureus DSP family
. Based on the venerable C6x floating point architecture, it supports true dual-zone decoding for multi-zone audio video recorders (AVRs) and enables better-sounding virtualizers or room simulators and high-quality room correction for commercial or automotive environments. The TMS320DA710 and TMS320DA708 processors are intended to support both decode and post-processing operations while the lower-priced TMS320DA707 and TMS320DA705 DSPs are intended to perform multi-channel decoding in applications where post-processing is handled separately. TI emphasizes its floating-point series because the extra processing power they pack allows developers to easily add extra value-add features and work with high-level tools like their Code Composer Studio
without having to worry about running out of MIPS.
For more cost-conscious applications, which do not require audio decoding, TI offers their 48-bit fixed point TAS3xxx Pure Path audio SoC platform
. Depending on which member of the PurePath family you choose, you can combine the 48-bit DSP core with several different combinations of analog input processing elements (ADCs, DACs and input muxes) and PWM controllers for digital audio outputs. All PurePath processors are supported by the PurePath Audio Development Environment which uses a graphic user interface to let you choose from a library of common audio functions and graphically connect them together. Included in the PurePath Studio software suite is the Integrated Development Environment, a traditional DSP code development environment consisting of an assembly code editor with contextual help facilities, an assembler, a simulator for debugging, and other tools.
Cirrus Logic's consumer audio offerings
are based on their Coyote 32-bit fixed-point core. The CS49xxx family
of dual core processors are typically used for surround sound decoders and other applications while the CS48xxx single core family
is intended for post-processing tasks in home and automotive audio. In late October Cirrus will introduce its CS47048 single core audio SOCs with integrated mixed signal elements including quad 32-bit 192 kHz sample rate ADCs, 32-bit DACs, and multiple channels of sample rate conversion capability.
Developers will be happy to know that Cirrus is moving away from a decade-long tradition of supplying almost all of the code for its products to offering basic building blocks and a full suite of user-friendly tools to assemble them with. Their composer suite provides a graphical DSP primitive programming environment that makes its low-level machine code assembler and high level C compiler easier to use. Cirrus’ library includes all the basic encode/decode and processing blocks but development capabilities enable developers to easily add their own secret sauce – such as an enhanced surround algorithm or software to get the most out of specific speaker/amplifier configurations.
Freescale audio offerings are centered around their new Symphony family of 24-bit DSPs
. Symphony processors feature a register-enabled double precision mode for operations where 48-bit accuracy is required. Freescale uses a 24-bit architecture (as opposed to TI’s 32-bits) because they think that it’s more cost-effective as most operations (around 80%) only require 24-bit operation to provide more than enough resolution. They say that invoking double-precision is only needed for things like processing lower frequencies where the extra bits are necessary to preserve and manage phase information.
The low end of the Freescale Symphony family includes single-core 100 MIPs devices
that are commonly used for sound enhancement in headphones, special effects in musical instruments, and decode-only functions in AV equipment. At the top end their dual-core 250 MHz devices can be found in everything from high-end AV amplifiers and guitar amps to mixing consoles. The dual-core chips have two independent processors but share key resources allowing inter-processor communication and load-sharing. Pin-compatibility across the whole product line allows OEMs to use a single design for several models with different feature sets.
Freescale has invested heavily in software, IP, and tools to make development faster and easier. All Symphony DSPs are supported by Symphony Studio, an Eclipse-compliant IDE that supports code creation and editing as well as project management, debugging and code compilation all in one software suite. The Symphony Studio includes a variety of enhancements including a C/C Development Tooling (CDT) plug-in for Eclipse that enables DSP software development in C and assembly. Debugging is achieved through the GNU project debugger interface within Eclipse. Symphony DSPs come with extensive libraries of ROM-based routines that includes all basic decoders and post-processing functions. Depending on your needs, there are versions available with licensed and license-free IP.Microchip
also plays heavily in consumer audio, although you may never see one of their lower-powered, lower-cost 16-bit dsPIC controllers
performing the sophisticated audio decode tasks that larger DSPs often do. True to their origins as a microcontroller maker, they entered the consumer audio from the embedded perspective rather than the Hi-Fi world.
Their products are designed to address the needs of embedded applications like speech compression/decompression, speech recognition/synthesis, sound synthesis, noise suppression, as well as acoustic- and line-echo cancellation. All the functions described above are available in Microchip’s library
of pre-packaged functions that can be downloaded for evaluation. Nearly all of the CEVA software is free to evaluate and only requires a one-time license fee if used in commercial applications.
Microchip's dsPIC 30F embedded signal controller delivers 30 MIPS worth of processing power and, depending on what you need, can be configured with 6 kbit – 144 kbit of Flash, and 256 bit – 8 kbit RAM in 18 - 80 pin packages. Some 30F devices include I2S and AC97 codec interfaces which allows connections to outside codecs that have more resolution than the integrated 12-bit cores. The dsPIC33 family offers more processing power(40 MIPS), more Flash (12 kbit – 256 kbit), and larger (1 kbit -30 kbit) RAM, plus 8 channels of transparent DMA using on-chip dual-port memory. Newer family members have dual 16-bit audio DACs (100 ksample/s): an industry first.Other Players
Time and space limitations (plus some company Failures to respond to our Editorial call) did not permit us to provide detailed analysis of the offerings of several other important players in the DSP market. We’ve provided basic information here to make sure you have all your options available to you when you decide to pick a DSP for your next audio design.CEVA
offers several audio DSP IP cores for use in ASICs and ASSPs, including their TEAK-Lite III
which targets a broad range of digital and home entertainment products such as HD audio, set-top boxes, digital TVs and audio receivers. It features native 32-bit processing and a dual multiply-accumulate (MAC) architecture, making the DSP suitable for deployment in High Definition (HD) audio applications requiring advanced audio standards such as Dolby Digital Plus 7.1, Dolby TrueHD and DTS-HD.STMicroelectronics
offers a range of DSPs for consumer audio products that range from simple enhancement and equalizers for boom boxes to full-blown decode/post-processor elements like their TDA7442 series
. For more information, visit their consumer audio page
- Lee H Goldberg