About VoiceApp

VoiceApp works by reading audio directly from your soundcard (in Linux it's file /dev/dsp), and transforming data via FFT. The result is thrown on the screen in matrixish colors ;).

mpg123 -s -m -4 shaggy-bombastik.mp3 ¦ ./voiceapp -w 600 -s (version 0.4.2)

Screenshots

Version 0.4.2 in action here(220k).

Requirements

Pentium (??) -- developed on PPro200/GF2PCI and eats 10%..
SDL (Simple DirectMedia Layer) -- www.libsdl.org.
FFTW (Fastest Fourier Transform in The West) -- www.fftw.org.
Linux (www.linux.com) -- VoiceApp sets some linux (OSS/Free??) specific settings, so porting it onto other platform is not easy. (as a side note, I have somewhere sources for same-kind app using DirectSound & MS Windows things..).
To get them under debian/unstable (don't know about stable):
apt-get install fftw-dev (needed to compile)
apt-get install sdl1.2-dev (-''-)
then make and ./voiceapp.

NOTE!
Please verify that your Microphone / line-in is not 'muted'.
VoiceApp needs some data input to work..

Download

Source:

• voiceapp-0.4.4.tar.gz
• Older version.. voiceapp-0.4.3.tgz
• Older version.. voiceapp-0.4.2.tgz
• Older version.. voiceapp-0.3.5.tgz
• Older version.. voiceapp-0.3.tgz.
• Older version.. voiceapp-0.2.tgz.

Fun things

• Power spectrum is shown after 'Bezier -curve' -fitting, use '-red', '-green' and '-blue' command line modifiers to produce more interesting colors.
• For example: ./voiceapp -red 0:0:0.2:1 -green 0:0:0:1 -blue 0:0.1:0.9:1 gives nice result.
• Try ./voiceapp -s < /dev/hda as root..
• Try whistling into microphone.
• Please do make strange voices ;)

ToDo

• Beat Detection (FFT again over power-sum)
• (done-0.3.5)ShiftingView
• ColorPresets
• [even] Better font handling (should be mappable directly to chars, so we can print anything we want..)
• more printable info.
• esound integration.
• improve FFT from a file. (slow-pace)
• pausing & snap-shotting.
• display of conventional tone when over some freq (440Hz = C₁)
• tons of things I've ever dreamt of =)
• directionality of new data.
• data interpolation on higher sample-counts. theoretically we have the horsepower, this would also soften out FFT-jungies..