/real-time-audio-fft

iOS library for analysing/visualising audio data at real-time

Primary LanguageObjective-CMIT LicenseMIT

Real-time audio analysis and FFT for iOS

This library helps you visualize or process sound as it is played by providing you with a currently-played spectrum data and samples, at any given time.

Features

  • Provides accurate and high-resolution data, partially written in C
  • Uses Apple's (fast) implementation of FFT - vDSP API from the Accelerate framework
  • Makes use of FFT overlapping (of 512 frames) for better timing
  • Can process real-time input from the microphone too

How to use

For playing audio files:

Initialization

self.audioFile = [[AudioFile alloc] init];
NSURL *url = [[NSBundle mainBundle] URLForResource:@"My Song" withExtension:@"mp3"];
[self.audioFile loadAudioWithURL:url withCompletionCallback:^(NSError *error)
{
   /* check error */
        
   // Let it play...
   error = [self.audioFile play];
        
   /* check error again */
}];

Getting live data

// Later in the program, for example in your update() function...
LiveAudioData realtimeAudioData = self.audioFile.liveAudioData;
LiveAudioChannelData leftChannel = realtimeAudioData.channel1;

if (leftChannel.containsData)
{
    // Get realtime frequencies data and waveform as pure float arrays
    float (*fftResults)[CHUNK_SIZE] = (float (*)[CHUNK_SIZE])leftChannel.fftResults.data;
    float *waveform = leftChannel.samples.data;
    UInt64 numFFTChunksAvailable = leftChannel.fftResults.numChunksAvailable;
    UInt64 numSamplesAvailable = leftChannel.samples.numSamplesAvailable;

    // Process the spectrum or the samples in various ways,
    // e.g drawing them to screen to create audio visualizations

    int frequency = 70;
    int bin = FrequencyToBinIndex(frequency, self.audioFile.playedAudioFormat.mSampleRate, CHUNK_SIZE);
    float bassMagnitudeInChannel1 = fftResults[0][bin];
    int height = bassMagnitudeInChannel1 / 2;
    self.band1.frame = CGRectMake(50, 500 - height, 15, height);

    // etc.
}