clap-detection
Clap sequence/rhythm/pattern detection on Raspberry Pi using Csound and Python. This was made as a quick experiment. Don't use it for anything serious.
Why
Some examples of things you can have your Raspberry Pi do when a matching clap sequence is detected:
- Dim the lights and start playing smooth jazz music
- Turn on/off the TV
- Broadcast a yo
- Project the weather forecast on the wall
How
Csound takes in audio from a microphone live and checks the audio for transients. Whenever a transient (rapidly ascending amplitude) is detected, Csound will notice ClapAnalyzer, a class implemented in Python. ClapAnalyzer looks for a specific rhytmic clap sequence. ClapAnalyzer will notice all listeners whenever a matching clap sequence is detected.
Setup
- Install Python 2.7
- Install Csound
Usage
Python
Let's use the following rhythmic sequence as example:
The note lengths are 1/4, 1/8, 1/8, 1/4, 1/4, so we set the note_lengths parameter to [0.25, 0.125, 0.125, 0.25, 0.25].
from clap import ClapAnalyzer
clap_analyzer = ClapAnalyzer(note_lengths=[0.25, 0.125, 0.125, 0.25, 0.25])
def clap_sequence_detected():
print 'Matching clap sequence detected!'
clap_analyzer.on_clap_sequence(clap_sequence_detected)
# You can now start calling clap_analyzer.clap(time)Basically, this is the python code that is used in clap.csd
Csound
Start csound from your command line. By default, the csound instrument will get live audio input:
csound clap.csd
If you want to quickly analyze a wav file, you can use that file instead of live audio input. This is good for testing:
csound clap.csd -i myfile.wav
PS: The file must be mono, not stereo, for this to work. And if your sound file is long, then you should modify the amount of time the Csound instrument stays alive accordingly, in order to analyze the whole file.
Troubleshooting
"no module named clap"
Try adding the directory with the python module dynamically:
pyruni "import sys, os"
pyruni "sys.path.append('/path/to/clap-detection')"
Edit /path/to/clap-detection to the place where clap.py is located.
"Segmentation fault" or "Unable to set number of channels on soundcard"
Check if your input device is mono or stereo. If it is mono (i.e. has only one channel), then you should set nchnls = 1 in your csound file, and you should use the in opcode instead of ins. If your input device is stereo, then you should set nchnls = 2.
ALSA and/or PortAudio warnings
Use the -+rtaudio=alsa option
Stuttering/crackling/noise/"Buffer underrun"
Let Csound use a large buffer in both software and hardware. In other word, use the following options: -b2048 -B2048
Input device error
Run arecord -l and check the list of sound cards and subdevices that are available. If you, for example, want to use card 1, subdevice 0, then you should use the following csound option: -i adc:hw:1,0