Audio Tools 0.8

cybero's picture

Great work getting the QuickTime Audio in and the Core Image Frequency Analyzer. Brilliant :-)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

cybero's picture
Re: Audio Tools 0.8

Probably a dumb question, but should one be able to load .mp3 from hypertext http URLs in Audio Tool paths?

Would it ever be feasible ?

cwright's picture
Re: Audio Tools 0.8

cybero wrote:
should one be able to load .mp3 from hypertext http URLs in Audio Tool paths?

Should, as in, should it work? no -- we only support file paths at the moment. Should, as in, should we get around to supporting URLs? yes.

cybero wrote:
Would it ever be feasible ?

Yep -- expect this functionality in 1.0 (probably out around the end of july/mid-august?)

(if you'd like this sooner, please contact us soon -- we're working on some other extensions to audiotools, so batching the work together would save a lot of time and effort).

cybero's picture
Re: Audio Tools 0.8

Thanks for the speedy confirmatory and advisory on the matter. Local paths only until further notice, then :-)

When URL loading becomes a reality, then the location and delivery of online audio and audio video quartz composer based presentations becomes possible.

Please find attached a work in progress, utilising your excellent Audio Image Filter. It is set to Live Input as I didn't want to burden anyone with one of my mix-downs KB-wise

PreviewAttachmentSize
frequencyspiral.qtz16.74 KB

cybero's picture
Re: Audio Tools 0.8

Works like a speedy little thing at lower scale settings. Makes for a great, deceptively simple, slimline little visualizer for iTunes. Set to live input, again.

PreviewAttachmentSize
frequencyspiral2.qtz17.04 KB

franz's picture
Re: Audio Tools 0.8

"When URL loading becomes a reality, then the location and delivery of online audio and audio video quartz composer based presentations becomes possible."

interesting. i see a great potential for museums installations, kiosks, corporate tv ....

dodo's picture
Re: Audio Tools 0.8

Great work and brillant idea having now access to the audio in stereo including the frequency analyser. I was waiting for such features for a while. I spent a lot of time building similar features using special settings in soundflower, several instances of LineIn and and then 2 Apple Audio input patches. The Kineme audio input patch is more powerful and the linear mode provides much more flexibility. This is the first release then using the new "audio-frequency.qtz" file I got some troubles on a MacMini: both Audio File Player and Audio File Loading patches fail. The Finder generates a huge bunch of warning messages in the Console related to the input connections and then fails to restore the patches operations. Last but not the least, the Audio input patch is working fine but the frequency analyser range is restricted to about 11kHz. I'm getting older but I still hear frequency above 11kHz. The Apple based spectrum analyser has 512 channels then 257 channels (including the useless channel 0) are necessary to cover the nominal 22kHz audio bandwidth. The patch is limited to 127 channels then 11kHz and this limit reflected back in all modes. Would it be feasible to extend the frequency range up to the nominal audio frequency bandwidth ? Please... a little effort...

Sorry, I'm not English native, thanks for your tolerance.

cwright's picture
Re: Audio Tools 0.8

dodo wrote:
This is the first release then using the new "audio-frequency.qtz" file I got some troubles on a MacMini: both Audio File Player and Audio File Loading patches fail. The Finder generates a huge bunch of warning messages in the Console related to the input connections and then fails to restore the patches operations.

Please post the console dump, and sample composition.

dodo wrote:
Last but not the least, the Audio input patch is working fine but the frequency analyser range is restricted to about 11kHz. I'm getting older but I still hear frequency above 11kHz. The Apple based spectrum analyser has 512 channels then 257 channels (including the useless channel 0) are necessary to cover the nominal 22kHz audio bandwidth. The patch is limited to 127 channels then 11kHz and this limit reflected back in all modes. Would it be feasible to extend the frequency range up to the nominal audio frequency bandwidth ?

This is incorrect -- For the Audio Input patch (live audio) We take the full 24khz (or 22.05khz, if you're in 44100Hz mode) spectrum, and divide it into 256 parts (when in linear mode -- all the other modes are derived from this 256-band mode). The number of output channels directly relates to the number of input samples -- we use 512 samples per channel; any more, and we're be behind by a frame, and that extra latency is bad. The next step up is 1024 samples (512 bands), which would only update 47 times per second -- not enough for 60fps animations. In some brief testing, I'm not able to whistle high enough to pass about 1/3 of the 127-band FFT on the sample composition, so I'm certain it handles all the way up to the Nyquist frequency. If you're not using the 127-band linear mode, you're going to miss a lot of the high end because all the other modes heavily bias the low frequencies (generally below 8-10khz, from my estimation).

The Audio File Input patch can use more samples (up to 65536 -- 32k bands, but very very slow) by adjusting the inputs on the patch. I don't think this is necessary for what you want though.

dodo's picture
Re: Audio Tools 0.8

Thanks for your prompt answer. Please find attached extracts from the console. The previous Audiotools v0.4 was OK. The "audio-frequency" composition sample is the one provided with the last audiotools release. I have to confess that I'm a bit confused regarding your explanation on the audio bandwidth behaviour. For instance, I have made some changes in "iTunes_LED_Spectrum_Analyser" (open source) using inputs from "iTunes Visual SDK" and the technical Note TN2016 all based on 512 channels / 44100Hz. I understand you most probably developped something new by your own and then you have my full respect. However I tested the Audio input patch with a bunch of audio test files, using soundflower to reroute the system audio output to the input, and I can confirm that the last displayed frequency is 11kHz and no more. The extra latency vs nb of channels is certainly a limitation but experiences have shown that the graphics load is very often the main driver regarding the fps degradation. Further Data processing is then necessary to reduce the amount of data to be displayed (javascript...). That's part of the user creativity considering some unavoidable trade-offs.

I'm very grateful to you for all your attention.

cwright's picture
Re: Audio Tools 0.8

All the warnings are from Finder -- it's complaining about making thumbnails, because AudioTools doesn't work in Safe Mode (Finder only generates icons in safemode). In otherwords, your finder icon will looks weird, but they're otherwise meaningless.

For the composition, hit space 3 times, to get to 127 band mode. That's the probably the only mode that has enough resolution to display high frequency data accurately. If this is the mode you're using already, can you please supply some high frequency audio files to test against? (steve did a bunch of signal processing junk that magically turns 512 samples of audio into 127 bands, not the expected 256... not sure what's happening there, but perhaps it's chopping at 1/2 Nyquist?)

dodo's picture
Re: Audio Tools 0.8

Finder Icons are fine then probably something else. One interesting symptom of the patch failure is the file path is not transmitted from the input splitter to the patch input. The path label at the output of the input splitter is blank as soon as connected to the Audio patch. The same for the frequency mode. If I disconnect any input splitter and then inject the data directly to the patch input, I can do it, but the patch doesn't work anyway. I need to start a deeper investigation at system level looking for some conflict with other running processes (or other QC plugin) but later (it's 3 am here).

Please find attached two audio test files among many others. That are linear frequency sweeps I use for simple calibration. The current frequency is determined using the sweep velocity respectively 10Hz/sec (20-2000Hz) and 100Hz/sec (2-20kHz). Reading the position in the file (or a synchronized clock) allows the current frequency to be displayed in the viewer if needed.

cwright's picture
Re: Audio Tools 0.8

dodo wrote:
One interesting symptom of the patch failure is the file path is not transmitted from the input splitter to the patch input. The path label at the output of the input splitter is blank as soon as connected to the Audio patch. The same for the frequency mode.

This is because of user error. QC wont propagate port values to unused patches. Since you're not using any of the outputs of the Audio File Input patch, QC doesn't bother sending the splitter values. If you toggle the Live Input in your composition, the Audio File Input patch gets values just fine. There is no error in this regard.

dodo wrote:
If I disconnect any input splitter and then inject the data directly to the patch input, I can do it, but the patch doesn't work anyway.

Again, this is because all the outputs from the patch aren't actually used -- they all go to a multiplexer that's using the other Kineme Audio patch (Audio Input).

dodo wrote:
Please find attached two audio test files among many others. That are linear frequency sweeps I use for simple calibration. The current frequency is determined using the sweep velocity respectively 10Hz/sec (20-2000Hz) and 100Hz/sec (2-20kHz). Reading the position in the file (or a synchronized clock) allows the current frequency to be displayed in the viewer if needed.

With these audio files, I can confirm it skipping the upper half of the frequency band -- I'll take a look and see why it's doing this. Thanks for the sample files!

cwright's picture
Re: Audio Tools 0.8

status update: I've isolated the frequency skipping problem, and have fixed it -- it will be in the next build of AudioTools. We were incorrectly discarding the top half of the frequency information.

gtoledo3's picture
Re: Audio Tools 0.8

lol... dude. I wish I would have looked at this thread, I would have given it a whirl with my test tone file. I hardly ever use spectrum... I wouldn't notice that in a million years.

This somehow brings to mind... Do you have to handle dithering explicitly, or does that just happen automatically... or is it not a concern at all with the AU setup?

FilipeQ's picture
Re: Audio Tools 0.8

Hi

It's possible to isolate a signal with certain frequency with this patch?

Thanks

smokris's picture
Re: Audio Tools 0.8

It's not really designed for precise signal detection. That said, if you don't require very good precision, it might be usable for that purpose.

Pick a frequency band that describes your signal using the Structure Index Member patch, and use the Conditional patch to test whether it exceeds a certain threshold.

FilipeQ's picture
Re: Audio Tools 0.8

First of all thanks for your reply. Sorry if this is a really dumb question, but how can i know the frequency of each band? And i also don't know how to do the comparison :S

Thanks

FilipeQ's picture
Re: Audio Tools 0.8

Hi I just started working with this patch, I'm trying to isolate the frequencies in the 20Hz to 1kHz band. Is it possible to accomplish that with this patch? And if yes can you guys give me some tips on how to do it. Thanks