home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!sgigate!odin!twilight!zuni!karsh
- From: karsh@sgi.com (Bruce Karsh)
- Newsgroups: comp.sys.sgi
- Subject: Re: Amesh source?
- Message-ID: <prnhbqk@zuni.esd.sgi.com>
- Date: 14 Sep 92 20:28:07 GMT
- References: <1992Sep14.112021.24762@u.washington.edu> <prjoui4@zola.esd.sgi.com>
- Sender: news@zuni.esd.sgi.com (Net News)
- Organization: Silicon Graphics, Inc.
- Lines: 22
-
- In article <prjoui4@zola.esd.sgi.com> portuesi@tweezers.esd.sgi.com (Michael Portuesi) writes:
-
- >The FFT is computed on the R3000 CPU. From a quick look
- >at the source, it looks like the FFT is being computed using
- >floating point arithmetic, off a pre-built table of sine and
- >cosine values. However, I'm not an expert on FFT code, so
- >I could be mistaken. Perhaps Bruce or one of the audio gurus
- >can correct me if I'm wrong.
-
- You are correct. Amesh is a perfectly normal AL/GL application. It
- uses a conventional single precision FFT running on the MIPS, not the
- DSP.
-
- >> My application is in speech signal processing and recognition; I
- >> would like to write a program which can, among other things, display
- >> a near-realtime spectrogram (2D, not 3D).
-
- This is well within the capabilities of the Indigo.
- --
-
- Bruce Karsh
- karsh@sgi.com
-