In June DSP core licensor CEVA announced CEVA-Audio, a licensable audio platform
targeting high-volume audio applications such as portable audio players and home
entertainment equipment. The CEVA-Audio platform is based on CEVA’s recently
announced CEVA-TeakLite-II core. For an overview of the CEVA-TeakLite-II, see
edition of Inside DSP.
The platform also includes a memory subsystem and peripherals. The size of
the memory is configurable, and includes up to 8 Kbytes of instruction cache
and 32 Kbytes of data memory. Peripherals include a DMA controller and a power
management unit. Licensees also have the option to include a TDM serial port
and an audio codec interface.
According to CEVA, the entire platform, including the core, memory, and peripherals,
occupies a die area of less than two square millimeters when fabricated in a
0.13-micron process. This is a notably small die area. For example, the entire
platform is about the same size as an ARM946E-S core without any memory.
CEVA claims that the platform is also highly energy-efficient, quoting 0.1
mA/MHz for the core and peripherals such as the DMA controller. BDTI’s
analysis shows that this is roughly comparable to other DSP cores targeting
low-cost audio applications, such as the LSI Logic ZSP400 and the Philips CoolFlux.
The CEVA-TeakLite-II at the heart of the platform achieves a worst-case clock
rate of 200 MHz when fabricated in the TSMC CL013G process. BDTI recently completed
an analysis of the computational requirements of audio algorithms. Based on
this analysis, the CEVA-Audio platform is likely fast enough for many mainstream
stereo and surround-sound applications.
Popular audio and speech codecs, including MP3, WMA, AAC, HE-AAC, Ogg Vorbis,
BSAC, NB-AMR and WB-AMR, are available or being developed for the platform.
Interestingly, CEVA is creating all of this software, rather than relying on
third parties. In the past, it was rare for processor vendors to create this
type of software themselves. However, there is a clear trend of vendors supplying
a growing amount of software, particularly for audio and video applications.
CEVA also provides an API for the platform that allows programmers to access
these codecs through simple commands on a host processor. Through this API,
the host processor can command the CEVA-Audio platform to perform pre-defined
functions such as changing songs or changing codecs. Licensees can also extend
this library with additional algorithms, use cases, and commands.
The CEVA-Audio platform is currently available for license. CEVA has not disclosed
the licensing fees or royalties for the platform.