Audio file compression as an effect?

VST, AU, AAX, CLAP, etc. Plugin Virtual Effects Discussion
RELATED
PRODUCTS

Post

Opinions?
Do not lick the fablanky

Post

Bitcrushers more or less do that do they not?

Post

for example in the film "the matrix", when neo takes the pill and is "revealed" as "digital form" or whatever is ment by that scene, they use a compression scheme very simmilar to that used by many mobile phone companies.

it involves analysis of the input signal to find frequency peaks in the signal, the inverse filter required to remove the peaks is applied, and the then "flattened" signal is encoded using very low depth ADPCM and sent with the filter commands. many systems also use generalized high frequency detection and detect common parameter changes and send them as more basic commands, like "A to B via slope #4" or etc.

the other side then does a "resynthesis" producing an aproximate copy of the input signal. you'll notice when talking on a mobile phone with poor reception, often an effect which sounds like the old bell labs voice synthesizer or the "neo" effect occurs. the system's bandwidth is dynamically adjusted to fit the bandwidth between the hub and the mobile device. during periods of lower bandwidth, the commands used for resynthesis are sent much less often. the effect produced is the equivalent of sample rate reduction, only on the resynthesis commands.

in short; yes, definitely. such techniques have already been used widely recently.
Last edited by aciddose on Thu Nov 03, 2005 1:03 am, edited 1 time in total.

Post

Reverse Engineer wrote:Bitcrushers more or less do that do they not?
They don't have the burbly noise of a cut-rate MP3 though.
Last edited by robenestobenz on Thu Nov 03, 2005 1:01 am, edited 1 time in total.

Post

not really, Mp3 files do a lot more than just reduce bitrate. frequency response is messed up and other stuff I don't really know thats why Im asking.

EDIT: referring to REverse Engineer, and wow, more posts!
Last edited by funkadil on Thu Nov 03, 2005 1:04 am, edited 1 time in total.
Do not lick the fablanky

Post

I've considered it as an effect before, but haven't got round to actually trying it out yet. There is actually a few spectral plugins which reproduce the effect pretty well.

Post

aciddose wrote:for example in the film "the matrix", when neo takes the pill and is "revealed" as "digital form" or whatever is ment by that scene, they use a compression scheme very simmilar to that used by many mobile phone companies.

it involves analysis of the input signal to find frequency peaks in the signal, the inverse filter required to remove the peaks is applied, and the then "flattened" signal is encoded using very low depth ADPCM and sent with the filter commands. many systems also use generalized high frequency detection and detect common parameter changes and send them as more basic commands, like "A to B via slope #4" or etc.

the other side then does a "resynthesis" producing an aproximate copy of the input signal. you'll notice when talking on a mobile phone with poor reception, often an effect which sounds like the old bell labs voice synthesizer or the "neo" effect occurs. the system's bandwidth is dynamically adjusted to fit the bandwidth between the hub and the mobile device. during periods of lower bandwidth, the commands used for resynthesis are sent much less often. the effect produced is the equivilant of sample rate reduction, only on the resynthesis commands.

in short; yes, definitely. such techniques have already been used widely recently.
Richard D James uses that effect a lot and I have always been jealous of it. It appears a lot in lo fi web streams.

I would LOVE such an effect. I've gotten the sound through noise reduction, but not by anything in real-time.

Post

Chase wrote:
aciddose wrote:for example in the film "the matrix", when neo takes the pill and is "revealed" as "digital form" or whatever is ment by that scene, they use a compression scheme very simmilar to that used by many mobile phone companies.

it involves analysis of the input signal to find frequency peaks in the signal, the inverse filter required to remove the peaks is applied, and the then "flattened" signal is encoded using very low depth ADPCM and sent with the filter commands. many systems also use generalized high frequency detection and detect common parameter changes and send them as more basic commands, like "A to B via slope #4" or etc.

the other side then does a "resynthesis" producing an aproximate copy of the input signal. you'll notice when talking on a mobile phone with poor reception, often an effect which sounds like the old bell labs voice synthesizer or the "neo" effect occurs. the system's bandwidth is dynamically adjusted to fit the bandwidth between the hub and the mobile device. during periods of lower bandwidth, the commands used for resynthesis are sent much less often. the effect produced is the equivilant of sample rate reduction, only on the resynthesis commands.

in short; yes, definitely. such techniques have already been used widely recently.
Richard D James uses that effect a lot and I have always been jealous of it. It appears a lot in lo fi web streams.

I would LOVE such an effect. I've gotten the sound through noise reduction, but not by anything in real-time.
Meatsynth does this, and I think convoluter. but Im too stingy to own either. can you do this in coagula and how? Ive never really figured that program out yet.
Do not lick the fablanky

Post

if you had a vst capable of routing to a directshow filter graph you could use all the codecs which come with for example 'msn messenger' or 'yahoo messenger' or etc. try 'graphedit' which came with the directx sdk. i'll upload, but notice, if you're from microsoft and this pisses you off, well, i'm sorry, upload it on your own server please.

http://xhip.cjb.net/temp/graphedit.zip

i could actually code such a vst with fair ease, however i dont have any of the required libs.

cant someone spend the time to do this?

dxi's are directshow plugins, i think, are they not?

Post

This would be a cool vst, yes. Speaking of which. I have this idea for a plugin which would "squarify" a wave. It would automatically set the level of the waveform to a certain level when it was positive and a certain level when it was negative. It would have a high-pass filter before it hits that component though so you could tweak it around and listen to the change. Good idea? Bad idea? It will probably sound bad just like my other ideas and have no use. :dog:
Do not lick the fablanky

Post

that can be done with a comparator (or hard-clipper with dc adjustment) and filter. perhaps i could add dc offset to my clipper, and a single filter? i'll let you know if i do. should take about 10 mins.

Post

funkadil wrote:This would be a cool vst, yes. Speaking of which. I have this idea for a plugin which would "squarify" a wave. It would automatically set the level of the waveform to a certain level when it was positive and a certain level when it was negative. It would have a high-pass filter before it hits that component though so you could tweak it around and listen to the change. Good idea? Bad idea? It will probably sound bad just like my other ideas and have no use. :dog:
i think PWMdrive does something like that.

Post

this is what mWarped Linear Prediction Filter (ok, not a good name, I know) was for. It inverse filters the input signal, quantizes the residual and then re-applies the spectral envelope (after frequency warping it, if so desired). It's a bit unstable and wasn't very well optimized, though, so it's perhaps a pain to use.

Post

actually that is the name of the process, LPC, linear predictive coding. i'm sure there are more peices of software out there too, not just one. the implementations i mentioned are actually not very good, though microsoft just updated thiers and it is much more efficent, its used in a new version of thier messenger application. since they started offering to act as a gateway for voice data, to compete with yahoo's service, i suppose they wanted to save bandwidth :)

Post

Weird, I thought that particular effect in The matrix can be done in a number of things, including Reaktor which has been cited several times being utilized in the sound effect work in that movie...

You can spot it done on a number of IDM things. Most of the artists in question are using, you guessed it... Reaktor :)

Check out any of the granular plugins, start with the selection from SmartElectronix, and you'll nail down that Matrix effect easily.

But yes, MP3 and other perceptual encoding schemes have been used as effects in music. Having such effects as plugins though, that'd be nice indeed.

Regards,

JMH
Now available with added Inherently Suspect Justification!

Post Reply

Return to “Effects”