| ??? 10/30/02 09:14 Read: times |
#31671 - RE: 8052/51 sound support |
Hey..........
You left this very same message 2 times in a thread yesterday... AND here you are starting yet another one on exactly the same subject. Two people had suggested you look at the ISD stuff...... I did a VERY simple search on www.google.com and found several things to look at that seem like worthwhile avenues to investigate. Here is one of the links. http://www.winbond-usa.com/products/...r/ If you don't like that one then I would go look at the chip offerings from a company called OKI SEMICONDUCTOR. Here is a chip specification link to a recorder / playback chip. http://www2.okisemi.com/us/docs/I...ml The reason that the others and myself are suggesting the use of these integrated solutions like this is that you get a degree of audio compression with these solutions. And thus it may make it practical for an 8051 to handle the data volume created when any amount of audio is to be buffered and stored. I have done much work on audio compression and decompression in a Video Conferencing product that I spent 6 years of my life working on. (Not all 6 years on the audio for sure). That product needed to reduce the bandwidth requirements for audio to a small amount so that it could be sent along with compressed video information through digital data link like ISDN phone link or ethernet. It is possible to compress audio down to a real time data flow of 8K bits per second or 4K bits per second in real time. You can learn about these compression techniques if you wish simply by doing search on the web for the terms "G.722" or "G.723" or "G.724". I would suggest that an 8051 is not up to the task of handling the algorithms needed to achieve audio compression to the 8K bits per second level. We used a DSP chip from TI running with a 60 MHz clock to do the job. And that DSP still used an Audio Codec chip to convert to/from analog audio to a digital audio format. The DSP would then compress and decompress that data. Note also that most people in business of making audio compression and decompression using the DSPs purchase the algorithms from one of several specialty software houses. This is because the software algoritnms are quite complicated. Now all that aside, even if you decide to look at one of these modular record/playback units you have to consider another aspect of working with audio data. If you follow the second link above you can find a very informative FAQ about the concepts behind the PCM and ADPCM sampling and voice encoding processes. It is possible to use a use an ADPCM sampling at 4KHz sample rate with audio and have a net data bandwidth of 16K bits per second (but believe me audio sampled at 4KHz sounds pretty bad). 16K bits per second is 2K bytes per second. At this rate you can hold a total of about 33 seconds of audio information in a 64K byte memory bank on an 8051. An 8051 can easily handle the data flow of the digitized ADPCM data but if you want to do any amount of significant audio storage you are going to have to put a good sized memory on the side of the 8051 to hold the data. I could comment that people often use DRAM of one flavor or another to hold audio data such as this. In days gone by when memory was much more expensive than it is today, it used to be possible to go to TI and others and purchase something called Audio DRAM that was normal DRAM memory chips that had known bad spots in it. They would assure that a given chip had a limited number of bad columns or rows in it. At about 25% the cost of fully functional DRAM it was a way to get to a larger sized audio store if you were willing to make a memory defect map and avoid the bad blocks. In todays world FLASH memory has changed everything. It is possible to purchase a quite large FLASH memory chip and get nice non-volatile storeage of audio data. For example, the seller DigiKey sells a series of FLASH parts called SerialDataFlash made by Atmel. You can get these in size from 1 Mbit to 64 M Bit. An 8 MBit part is in the $8->$12 US price and would hold 524 seconds of ADPCM data sampled at 4KHz rate. If you went with the way better sounding 8KHz sampling rate you could expect to hold 262 seconds of audio in a serial flash like that. That is about 4.3 minutes. (Similar to the storage you can expect in a lower cost digital telephone answering machie!!). Serial flash is also quite easy to connect to an 8051. Best of luck... Mike Karas |
| Topic | Author | Date |
| 8052 Voice encoder/decoder | 01/01/70 00:00 | |
| RE: 8052 Voice encoder/decoder | 01/01/70 00:00 | |
| RE: 8052 Voice encoder/decoder | 01/01/70 00:00 | |
| RE: 8052/51 sound support | 01/01/70 00:00 | |
RE: 8052/51 sound support | 01/01/70 00:00 |



