Building a visualizer: Sound and signal processing - Developers

Howdy, I want to take a stab at creating a pimp'n visualizer. To do that I need to be able to do some spectrum analyses on a given song. As I do not have any experience with signal processing, I was hoping those of you who do could point me in the right direction. What tools/libraries should I use if I want to be able to detect pitch and tonal changes as well as volume and hopefully BPM. I'm comfortable using any modern language. PS: I don't expect this to be a real-time application.

Slightly off topic. But has anyone ever considered using image recognition on a visualiser, as a form of sound recognition.

[QUOTE=hcross1;46994272]Slightly off topic. But has anyone ever considered using image recognition on a visualiser, as a form of sound recognition.[/QUOTE] what so like using the visual translation of the wave movements to recognize "HEY THATS A BASS DROP I CAN TELL BECAUSE IT LOOKS LIKE ONE" that would be cool

[QUOTE=No Party Hats;46994303]what so like using the visual translation of the wave movements to recognize "HEY THATS A BASS DROP I CAN TELL BECAUSE IT LOOKS LIKE ONE" that would be cool[/QUOTE] I was once working on a project which involved a LOT of voice editing work, which involved spending hours looking at visual translations of words. Over a long period of time I was actually able to recognise basic words just from their visualisation. Computers could do this too, similar to how face recognition technology works.

[QUOTE=hcross1;46994272]Slightly off topic. But has anyone ever considered using image recognition on a visualiser, as a form of sound recognition.[/QUOTE] You could do, but there wouldn't be any benefit over using traditional signal processing techniques (and depending on how the visualiser works there's probably be drawbacks). It's a cool idea though! I don't know much about relevant libraries etc but if you need any general help regarding signal processing techniques feel free to pm me and I'll try to give you a hand.

OP, look at the convolution, there might lie your answer

Well believe it or not, sound files contain all the data you need. If you take the bytes of a wav file, and map them across the x-axis, the amplitude of the bytes will actually reflect the sound wave across the x-axis. I.e:[url=http://facepunch.com/showthread.php?t=1439640&p=46624543&viewfull=1#post46624543]Unless you want to display some other kind of data, I believe this is what you are looking for.[/url] Now, I did mine a stupid way. I dumped the data into a text file and read it through that (but that's because I was doing it in basic and have no idea how reading binary files in basic works). If you did this in another language, say C++, there are standard library functions to directly read binary data that you can map as [b]SIGNED[/b] 16-bit integers (signed is very important; I forgot this)

[QUOTE=proboardslol;46995527]Well believe it or not, sound files contain all the data you need. If you take the bytes of a wav file, and map them across the x-axis, the amplitude of the bytes will actually reflect the sound wave across the x-axis. I.e:[url=http://facepunch.com/showthread.php?t=1439640&p=46624543&viewfull=1#post46624543]Unless you want to display some other kind of data, I believe this is what you are looking for.[/url] Now, I did mine a stupid way. I dumped the data into a text file and read it through that (but that's because I was doing it in basic and have no idea how reading binary files in basic works). If you did this in another language, say C++, there are standard library functions to directly read binary data that you can map as [b]SIGNED[/b] 16-bit integers (signed is very important; I forgot this)[/QUOTE] Ohh snap, that is almost exactly what I was looking for! This might be easier than I originally thought.

So I'm now creating a visualizer with the data that I'm pulling. Need to name this project, any ideas? Naming things is by far the hardest part of programming.