Channels - spadpcm1 by Onslaught

Channels - spadpcm1 by Onslaught [web]

screenshot added by algorithm on 2013-11-30 00:36:41

platform :	Commodore 64 Commodore 64
type :	demo demo
release date :	november 2013

popularity : 52%

52%

0.80

alltime top: #20960

algorithm [code]
CONS [graphics (picture)]

added on the 2013-11-30 00:36:41 by algorithm

popularity helper

increase the popularity of this prod by spreading this URL:

or via: facebook twitter pinterest tumblr bluesky threads

comments

A single filer demonstrating the output of my sp-adpcm1 encoder that features multiple prediction and more accurate block analysis.

Decoding and resequencing done in realtime on the c64 at 7812 samples per second with ample processing time left for other effects

added on the 2013-11-30 00:37:37 by algorithm

Very nice! Lossy LZ used in this one?

rulez added on the 2013-11-30 01:56:06 by trixter

Its my implementation of step CVSD (or SP-ADPCM) and the decoder can also play loops while decoding allowing more overall compression. Decode is not optimised (could have optimised it two fold) but still plays back at nearly 8khz leaving half the processing time for other things

added on the 2013-11-30 02:17:16 by algorithm

I cant get it to run in Vice. :(

added on the 2013-11-30 04:45:39 by tonyrocks

Worked on Vice here, v2.4
Holy shit, it's sound great. I liked to hear a more complex tune with accoutic intruments like piano.
"half the processing time for other things" yet
fuck yes!

rulez added on the 2013-11-30 11:12:06 by Buckethead

and all on a little .prg <3 2min! (but the nice music feel repetitive)

added on the 2013-11-30 11:18:27 by Buckethead

Impressive!!

rulez added on the 2013-11-30 12:05:05 by BackSpace

Indeed. I like the picture too.

rulez added on the 2013-11-30 12:56:33 by Emod

YOU are a genius! Welcome to the club =)

rulez added on the 2013-11-30 14:32:16 by vibrator

Hm, it does sound a little repetitive... but I guess it's a significant achievement anyway. :)

added on the 2013-11-30 14:34:19 by vibrator

Uncompressed, the audio would use approximately a megabyte (8bit 7812hz). Removing the repeating patterns and resequencing would still take up around 300k of data. The 1bit compression compresses this data to around 40k or so. Each 4bar pattern uses 2k of data (expanded to around 16k)

The decoder is not optimised at all, but still uses some type of direct nibble to code method that directly decodes each 4bit bitstream to a specific code that decodes the 4 samples (could have opted for the direct8bit to 8 samples, but that would have required more code tables.

added on the 2013-11-30 16:18:33 by algorithm

nice

rulez added on the 2013-11-30 17:40:30 by rudi