Dear Forum.
I need to process a Audio signal of 16bit /20Khz circa,Encode in MP3 and Transmit
The goal of the project is to consume less power than possible.
I have a little experience with DSP from Texas,the c6000 family, so,question are:
1)are a good choice to develop mp3 via sw, using a lot of library on the web (TI provide his DSPLIB),or,
is better to use a MP3 encoder .via some asic chip
2)Are performance of MP3 sw good as chip, or speed limit some quality.
Any suggestion already tested ,with some data, could be good
Thanks for your time
Roberto
Roberto-
Unless you need to combine MP3 with other audio processing, probably the ASIC approach will give you the lowest cost, footprint, and power consumption. For 20 years MP3 has been a standardized thing, so commercially available chips should be fully compatible and reliable.
TI is no longer pushing a future DSP roadmap so that's another thing to consider.
-Jeff
Thanks Mr Jeff for your reply
Im very disappointed that TI dont consider DSP more so important..also because Im learnin them.
Do you have a idea ofwhy ? What alternative (maybe SoC)?
Thanks
Roberto
Roberto-
TI ARM+DSP based SoCs and other vendor (FPGA) SoCs continue to be highly effective solutions to a wide range of problems. But again, in your case, they would likely be overly expensive and complex if you only need to do MP3.
Regarding the current state of commercial DSP and TI's roadmap (or lack thereof), it's not good. In 2015 TI had a great opportunity to move their advanced DSPs, which by that time had morphed into multicore CPUs, into deep learning and AI markets. However, that would have required them to engage in servers, PCIe cards, open source, and deep learning R&D -- substantial, material changes from their traditional world view that "everyone starts their project with an EVM board and a JTAG emulator". Unfortunately, the world has moved on from that approach, and somehow TI just couldn't see it.
I wrote a post with more background on what happened here Are DSPs obsolete ?
-Jeff
