How to write audio file locally recorded from microphone using AudioBuffer in iPhone?

Question

I am new to Audio framework, anyone help me to write the audio file which is playing by capturing from microphone?

below is the code to play mic input through iphone speaker, now i would like to save the audio in iphone for future use.

i found the code from here to record audio using microphone http://www.stefanpopp.de/2011/capture-iphone-microphone/

/**

Code start from here for playing the recorded voice 

*/

static OSStatus playbackCallback(void *inRefCon, 
                                 AudioUnitRenderActionFlags *ioActionFlags, 
                                 const AudioTimeStamp *inTimeStamp, 
                                 UInt32 inBusNumber, 
                                 UInt32 inNumberFrames, 
                                 AudioBufferList *ioData) {    

    /**
     This is the reference to the object who owns the callback.
     */
    AudioProcessor *audioProcessor = (AudioProcessor*) inRefCon;

    // iterate over incoming stream an copy to output stream
    for (int i=0; i < ioData->mNumberBuffers; i++) { 
        AudioBuffer buffer = ioData->mBuffers[i];

        // find minimum size
        UInt32 size = min(buffer.mDataByteSize, [audioProcessor audioBuffer].mDataByteSize);

        // copy buffer to audio buffer which gets played after function return
        memcpy(buffer.mData, [audioProcessor audioBuffer].mData, size);

        // set data size
        buffer.mDataByteSize = size; 

         // get a pointer to the recorder struct variable
Recorder recInfo = audioProcessor.audioRecorder;
// write the bytes
OSStatus audioErr = noErr;
if (recInfo.running) {
    audioErr = AudioFileWriteBytes (recInfo.recordFile,
                                    false,
                                    recInfo.inStartingByte,
                                    &size,
                                    &buffer.mData);
    assert (audioErr == noErr);
    // increment our byte count
    recInfo.inStartingByte += (SInt64)size;// size should be number of bytes
    audioProcessor.audioRecorder = recInfo;

     }
    }

    return noErr;
}

-(void)prepareAudioFileToRecord{

NSArray *paths =             NSSearchPathForDirectoriesInDomains(NSDocumentDirectory,NSUserDomainMask, YES);
NSString *basePath = ([paths count] > 0) ? [paths objectAtIndex:0] : nil;

NSTimeInterval time = ([[NSDate date] timeIntervalSince1970]); // returned as a double
long digits = (long)time; // this is the first 10 digits
int decimalDigits = (int)(fmod(time, 1) * 1000); // this will get the 3 missing digits
//    long timestamp = (digits * 1000) + decimalDigits;
NSString *timeStampValue = [NSString stringWithFormat:@"%ld",digits];
//    NSString *timeStampValue = [NSString stringWithFormat:@"%ld.%d",digits ,decimalDigits];


NSString *fileName = [NSString stringWithFormat:@"test%@.caf",timeStampValue];
NSString *filePath = [basePath stringByAppendingPathComponent:fileName];
NSURL *fileURL = [NSURL fileURLWithPath:filePath];
// modify the ASBD (see EDIT: towards the end of this post!)
audioFormat.mFormatFlags = kAudioFormatFlagIsBigEndian | kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked;

// set up the file (bridge cast will differ if using ARC)
OSStatus audioErr = noErr;
audioErr = AudioFileCreateWithURL((CFURLRef)fileURL,
                                  kAudioFileCAFType,
                                  &audioFormat,
                                  kAudioFileFlags_EraseFile,
                                  &audioRecorder.recordFile);


assert (audioErr == noErr);// simple error checking
audioRecorder.inStartingByte = 0;
audioRecorder.running = true;
self.audioRecorder = audioRecorder;

}

thanks in advance bala

Colyer

If possible, you can give a try to AVAudioRecorder, which makes it simple.techotopia.com/index.php/…. – Brinker

Can you post the code for your updated playbackCallback function and also the code that sets up the Audio Stream Basic Description. That would be a great help thanks – Urea

I updated answer. Changed the property added, and also you should delete the line of code in prepareAudioFileToRecord method that adds the kAudioFormatFlagIsBigEndian to audioFormat.mFormatFlags and move it to within the -(void)initializeAudio method. – Urea

Answer 1

To write the bytes from an AudioBuffer to a file locally we need the help from the AudioFileServices link class which is included in the AudioToolbox framework.

Conceptually we will do the following - set up an audio file and maintain a reference to it (we need this reference to be accessible from the render callback that you included in your post). We also need to keep track of the number of bytes that are written for each time the callback is called. Finally a flag to check that will let us know to stop writing to file and close the file.

Because the code in the link you provided declares an AudioStreamBasicDescription which is LPCM and hence constant bit rate, we can use the AudioFileWriteBytes function (writing compressed audio is more involved and would use AudioFileWritePackets function instead).

Let's start by declaring a custom struct (which contains all the extra data we'll need) and adding an instance variable of this custom struct and also making a property that points to the struct variable. We'll add this to the AudioProcessor custom class, as you already have access to this object from within the callback where you typecast in this line.

AudioProcessor *audioProcessor = (AudioProcessor*) inRefCon;

Add this to AudioProcessor.h (above the @interface)

typedef struct Recorder {
AudioFileID recordFile;
SInt64 inStartingByte;
Boolean running;
} Recorder;

Now let's add an instance variable and also make it a pointer property and assign it to the instance variable (so we can access it from within the callback function). In the @interface add an instance variable named audioRecorder and also make the ASBD available to the class.

Recorder audioRecorder;
AudioStreamBasicDescription recordFormat;// assign this ivar to where the asbd is created in the class

In the method -(void)initializeAudio comment out or delete this line as we have made recordFormat an ivar.

//AudioStreamBasicDescription recordFormat;

Now add the kAudioFormatFlagIsBigEndian format flag to where the ASBD is set up.

// also modify the ASBD in the AudioProcessor classes -(void)initializeAudio method (see EDIT: towards the end of this post!)
    recordFormat.mFormatFlags = kAudioFormatFlagIsBigEndian | kAudioFormatFlagIsSignedInteger | kAudioFormatFlagIsPacked;

And finally add it as a property that is a pointer to the audioRecorder instance variable and don't forget to synthesise it in AudioProcessor.m. We will name the pointer property audioRecorderPointer

@property Recorder *audioRecorderPointer;

// in .m synthesise the property
@synthesize audioRecorderPointer;

Now let's assign the pointer to the ivar (this could be placed in the -(void)initializeAudio method of the AudioProcessor class)

// ASSIGN POINTER PROPERTY TO IVAR
self.audioRecorderPointer = &audioRecorder;

Now in the AudioProcessor.m let's add a method to setup the file and open it so we can write to it. This should be called before you start the AUGraph running.

-(void)prepareAudioFileToRecord {
// lets set up a test file in the documents directory
    NSArray *paths = NSSearchPathForDirectoriesInDomains(NSDocumentDirectory,NSUserDomainMask, YES);
    NSString *basePath = ([paths count] > 0) ? [paths objectAtIndex:0] : nil;
    NSString *fileName = @"test_recording.aif";
    NSString *filePath = [basePath stringByAppendingPathComponent:fileName];
    NSURL *fileURL = [NSURL fileURLWithPath:filePath];

// set up the file (bridge cast will differ if using ARC)
OSStatus audioErr = noErr;
audioErr = AudioFileCreateWithURL((CFURLRef)fileURL,
                                  kAudioFileAIFFType,
                                  recordFormat,
                                  kAudioFileFlags_EraseFile,
                                  &audioRecorder.recordFile);
assert (audioErr == noErr);// simple error checking
audioRecorder.inStartingByte = 0;
audioRecorder.running = true;
}

Okay, we are nearly there. Now we have a file to write to, and an AudioFileID that can be accessed from the render callback. So inside the callback function you posted add the following right before you return noErr at the end of the method.

// get a pointer to the recorder struct instance variable
Recorder *recInfo = audioProcessor.audioRecorderPointer;
// write the bytes
OSStatus audioErr = noErr;
if (recInfo->running) {
audioErr = AudioFileWriteBytes (recInfo->recordFile,
                                false,
                                recInfo->inStartingByte,
                                &size,
                                buffer.mData);
assert (audioErr == noErr);
// increment our byte count
recInfo->inStartingByte += (SInt64)size;// size should be number of bytes
}

When we want to stop recording (probably invoked by some user action), simply make the running boolean false and close the file like this somewhere in the AudioProcessor class.

audioRecorder.running = false;
OSStatus audioErr = AudioFileClose(audioRecorder.recordFile);
assert (audioErr == noErr);

EDIT: the endianness of the samples need to be big endian for the file so add the kAudioFormatFlagIsBigEndian bit mask flag to the ASBD in the source code found at the link provided in question.

For extra info about this topic the Apple documents are a great resource and I also recommend reading 'Learning Core Audio' by Chris Adamson and Kevin Avila (of which I own a copy).

Urea

Awesome, i can able to save the file but i can't play it, is there anything issue with the format? – Prejudge

Hey Bamsworld, your solution is working good for saving the audio loclay, but can you tell me the right settings to save the audio clearly (less background noise) and playable format? right now the saved file is not playable. – Colyer

@Prejudge try replacing kAudioFileCAFType with kAudioFileAIFFType in the AudioFileCreateWithURL function call. Also add the format flag kAudioFormatFlagIsBigEndian to the AudioStreamBasicDescription where it is created in source code from link provided. I will update answer. – Urea

@Colyer see comments above and updated answer. I suspect the endianess of the samples has something to do with this. Also I had the audio file type as Core Audio File Format (caf) where we want Audio Interchange File Format (aiff), and MUST be big endian. I hope this helps. – Urea

I did the changes but still the saved files are not playable. also all the file sizes are 6144 bytes. it seems like the saving is working wrongly. – Colyer

@Urea Can you help me to save the complete audio file? now its seems like overwriting all the packets. – Colyer

@Urea I added audioProcessor.audioRecorder = recInfo after this line recInfo.inStartingByte += (SInt64)size, now i am getting a lengthy file but when i play it. getting 'urrrrrr' sound. waiting for your solution – Colyer

@Prejudge I have updated answer and tested the code and it all works as it should. The main issue was where we write bytes, I originally posted &buffer.mdata where it should be buffer.mdata as this is already a pointer of type void. Also corrected the property to be a pointer to the struct variable. I hope this helps you. – Urea

@Urea thanks your solution works for ".aif" and ".caf" codec but how to make compatible for ".wav" i changed recorder format in prepare method but not works. also not getting how to fetch NSData in recording callback. – Lonesome

Answer 2

Use Audio Queue Services.

There is an example in the Apple documentation that does exactly what you ask:

Audio Queue Services Programming Guide - Recording Audio

Physical

I can able to record the audio and play it through the iPhone speaker lively , my question is how to save the recording audio into local directory as a audio file? – Colyer

@Colyer see the link. It describes recording to a file. – Physical