Hello list.....Intrigued by the idea of playing back jpeg (or other) files as audio signals I've been having a go with the readsf~ and soundfiler objects....skipping the header with readsf~ and using the -raw flag in soundfiler, opening in an array and using tabplay~ to play it back....whilst I can get these objects to open the files and play them back, the results seem a bit disappointing...just a lot of white noise at a constant volume. The wave forms displayed in the array look more promising than the audio result Was I expecting too much? What else could I be doing to achieve a more interesting sound? Nick.
nick burge a écrit :
Hello list.....Intrigued by the idea of playing back jpeg (or other) files as audio signals I've been having a go with the readsf~ and soundfiler objects....skipping the header with readsf~ and using the -raw flag in soundfiler, opening in an array and using tabplay~ to play it back....whilst I can get these objects to open the files and play them back, the results seem a bit disappointing...just a lot of white noise at a constant volume.
yep, images are 8 bits, so S/N should be around 50dB.
The wave forms displayed in the array look more promising than the audio result Was I expecting too much? What else could I be doing to achieve a more interesting sound?
using high bit depth images (???) c
Nick.
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
cyrille henry wrote:
yep, images are 8 bits, so S/N should be around 50dB.
this only holds true if you are know about this when loading the file. since the files are opened as "raw", they will be interpreted as 16bit or higher.
fgasdr IOhannes
nick burge wrote:
Hello list.....Intrigued by the idea of playing back jpeg (or other) files as audio signals I've been having a go with the readsf~ and soundfiler objects....skipping the header with readsf~ and using the -raw flag in soundfiler, opening in an array and using tabplay~ to play it back....whilst I can get these objects to open the files and play them back, the results seem a bit disappointing...just a lot of white noise at a constant volume. The wave forms displayed in the array look more promising than the audio result Was I expecting too much? What else could I be doing to achieve a more interesting sound?
Jpegs are usually compressed and compressed data looks like white noise. Maybe try uncompressed images with regular patterns in them, or low-pass filter the sound.
Martin
On Sun, 2009-07-05 at 12:46 -0400, Martin Peach wrote:
nick burge wrote:
Hello list.....Intrigued by the idea of playing back jpeg (or other) files as audio signals I've been having a go with the readsf~ and soundfiler objects....skipping the header with readsf~ and using the -raw flag in soundfiler, opening in an array and using tabplay~ to play it back....whilst I can get these objects to open the files and play them back, the results seem a bit disappointing...just a lot of white noise at a constant volume. The wave forms displayed in the array look more promising than the audio result Was I expecting too much? What else could I be doing to achieve a more interesting sound?
Jpegs are usually compressed and compressed data looks like white noise. Maybe try uncompressed images with regular patterns in them, or low-pass filter the sound.
or try single-channel files (black&white) instead of coloured files.
depending on how you read the files and what bitdepth you assume (8/16/24/32bit), the same files is going to sound much different. let's say, you read a plain 3-channel rgb uncompressed image file (uncompressed tiff or bmp), where each channel has a bit depth of 8 bit, and you open such a file, while specifying 24 bit per sample, then your hearing sensation would almost only be influenced by the most significant byte (so to speak: by one of the three color channels). in order to treat all the channels the same, you would need to extract each channels first.
as a start, it might be easier to play with only black and white images. the problem here is, that you cannot tell [readsf~] to assume 8 bit input. however, you could read it as 16 bit input and then extract the lower and upper 8 bit from each sample and write the resulting two samples to another array.
rendered images containing simple gradients might sound less like pure noise than real photos. due to how many image formats are encoded, the width in pixel might correspond with the base frequency of the resuling sound, so this might be an important parameter to play with as well.
yo, tell us, if you can come up with something 'interesting'. i'd be interested to hear and see.
roman
Telefonate ohne weitere Kosten vom PC zum PC: http://messenger.yahoo.de
-
cyrille henry -
IOhannes m zmoelnig -
Martin Peach -
nick burge -
Roman Haefeli