home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.sys.amiga.advocacy:29809 comp.sys.atari.st.tech:5784
- Newsgroups: comp.sys.amiga.advocacy,comp.sys.atari.st.tech
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!hanauma.jpl.nasa.gov!hyc
- From: hyc@hanauma.jpl.nasa.gov (Howard Chu)
- Subject: Re: DSP not as good as we thought?
- Message-ID: <1992Nov24.020031.10135@elroy.jpl.nasa.gov>
- Sender: news@elroy.jpl.nasa.gov (Usenet)
- Nntp-Posting-Host: hanauma.jpl.nasa.gov
- Organization: SAR Systems Development & Processing, JPL
- References: <1992Nov20.184901.15161@dcs.warwick.ac.uk> <37211@cbmvax.commodore.com>
- Date: Tue, 24 Nov 1992 02:00:31 GMT
- Lines: 59
-
- In article <37211@cbmvax.commodore.com> grr@cbmvax.commodore.com (George Robbins) writes:
- >In article <1992Nov20.184901.15161@dcs.warwick.ac.uk> mashton@dcs.warwick.ac.uk (M.S. Ashton) writes:
- >> Am I to understand (from reading a review of the Falcon) that the DSP chip
- >> is not capable of reproducing digital sound at variable frequency (like Paula
- >> can) and is, in fact, limited to fixed rates such as 12.5,25,50KHz.
- >> If this is the case then all the nice DMA sound is completely unusable for
- >> computer generated music (soundtracker/Med etc.) and only useful from replaying
- >> a sample at the frequency it was sampled at.
- >> Am I to understand that to get round this, one needs to use masses of
- >
- >The DSP *should* have to computational capabilites to do real-time
- >sample rate conversion and filtering as it "plays" the sounds.
- >
- >The variable sample rate in the Amiga is one of those things that
- >is intended to minimize memory/computational requirements when
- >used with a simple DAC scheme, but given a farily powerful DSP,
- >there are other ways of getting the same job done.
-
- The sound hardware in the Falcon can use one of three different clocks - 32MHz,
- 25 MHz, or external clock input. Additionally, the clock can be divided down
- by one of about 16 different standard prescale values. This is what's given in
- the standard sound system calls. However, it seems that the built-in codec is
- only allowed to be used with the 25 MHz clock. I don't know why that is...
- It's not a problem for the DSP, though. Regardless, the codec supports more
- speeds than just 12.5, 25 and 50 khz, those are just the common speeds that
- were also supported on the STe.
-
- As long as this is going to be a topic of discussion, please remember that
- there are several independent elements of the Falcon audio system that can
- be interconnected in a variety of ways. You can use all, some, or none of them
- as you see fit. In fact there are so many different elements available it's
- difficult to choose where to start, in a system description.
-
- You have the codec with stereo 16-bit ADC and DAC, DMA record channels, DMA
- playback channels, external inputs and outputs, etc. The codec's ADC can be
- connected to the mic input or to the Yamaha PSG output. (Independent left and
- right channel control there.) You can independently activate any of the 4
- stereo audio tracks, and select any one of those tracks to be monitored by
- the internal speaker/DAC/stereo headphone jack. You can record or playback
- in any of 8-bit mono, 8-bit stereo, or 16-bit stereo. At system bootup I
- believe the ADC gets both channels from the PSG, and everything else is
- bypassed. For the voicemail software that I wrote (that was running all week
- at Comdex) on the Falcon, I had the audio matrix connecting the ADC to the
- DMA record channels, using only a single track. For the next version I'll have
- the ADC feeding the DSP, do some compression in the DSP, and feed the DSP
- output to the DMA record channel instead. The Falcon audio system is
- incredibly flexible, I only need to add two system calls to my existing code
- to add this functionality. (Oh, and load my compression routine into the DSP,
- but that's really a separate issue. That's the total impact on my code, tho.)
-
- One of the quirks I've noted is that the PSG doesn't have a hardwired
- connection to the speaker, and the system bell and keyclick are still
- generated there. If you have sound software that wants to record thru the
- microphone input, those system sounds disappear (unless you only set one
- channel to the mic, and leave one channel for the PSG...).
- --
- -- Howard Chu @ Jet Propulsion Laboratory, Pasadena, CA
-
- All true wisdom is conveyed in one-line witticisms.
-