dart-mic

Who are we?

We are Max Deibel (@mdeibel, [email protected]) and Patrick Yukman (@pyukman, [email protected]). We created this library as a stand-alone project, but it was motivated by our work on our senior project at Dartmouth College: VoxTrainer.

About the library

This is a Javascript library which listens to microphone input and performs pitch/note detection, volume detection, recording, and general purpose data processing. It makes use of the Web Audio APi (which is only supported by Chrome currently) and DSP.js.

To view a website with a simple demo of our project, please visit here: DEMO

General Usage

1. Create Microphone object (var mic = new Microphone();).
2. Any function defined below as

   this.foo = function(params) { ... }

   can then be called via the Microphone object. Example:

   mic.foo(param);

Function list (see "Function Descriptions" for more details)

Microphone Setup

Callable functions	Helper Functions
this.initialize	gotStream
this.isInitialized	noStream
this.startListening
this.stopListening
this.startRecording
this.stopRecording

General Purpose

Callable functions	Helper Functions
this.getMaxInputAmplitude	matchNote
this.getFreq
this.getNote
this.getNoteCents

Autocorrelation

Callable functions	Helper Functions
autocorrelate
getPeakPeriodicity
computeFreqFromAutocorr
getNoteFromAutocorr
getNoteCentsFromAutocorr

FFT

Callable functions	Helper Functions
	computeFreqFromFFT
	jainsMethodInterpolate
	getNoteFromFFT
	getNoteCentsFromFFT

Recording

Callable functions	Helper Functions
	writeToWav
	interleave
	mergeBuffers
	writeUTFBytes

Testing and Debugging

Callable Functions	Helper Functions
this.logData	destroyClickedElement

Function descriptions

Microphone Setup Functions

this.initialize Properly initializes the parameters of a Microphone object, defines the frequency-->note lookup table, and calls getUserMedia to request browser-level access to a user's microphone. In general, do not change any part of this initialize function without a compelling reason. Example: mic.initialize()

this.isInitialized This function simply returns whether or not the microphone object has been fully initialized (indicated by the var initialized being equal to true. Returns a boolean value. Example: if (mic.isInitialized()) { ... }

this.startListening Connects the microphone input to a processing node for future operations. Throws an error if the microphone hasn't been initialized before this function is called -- in other words, if a user tries to get mic data before allowing the browser permission to collect it. Example: mic.startListening()

this.stopListening Disconnects the microphone input. Can be called or tied to a button to save on processing. Example: mic.stopListening()

this.startRecording Begins gathering microphone input and storing it in a WAV file. Example: mic.startRecording()

this.stopRecording Stops packaging incoming microphone data into WAV file. Example: mic.stopRecording()

gotStream This function is the success callback for getUserMedia and initializes the Web Audio API / DSP.JS structures that allow us to manipulate the data streaming in off the microphone.

noStream This function is the failure callback for getUserMedia and alerts the user if their browser doesn't support getUserMedia.

General purpose functions

this.getMaxInputAmplitude Input: none. Output: the amplitude of the microphone signal, expressed in decibels (scaled from -120). Gives an idea of the volume of the sound picked up by the microphone. Example: mic.getMaxInputAmplitude()

this.getFreq Input: the method number (default is 1 to use autocorrelation, 2 to use FFT). Output: the detected frequency calculated via the selected method. Example: mic.getFreq(1) returns frequency calculated via autocorrelation

this.getNote Input: the method number (default is 1 to use autocorrelation, 2 to use FFT). Output: the detected note calculated via the selected method. Example: mic.getNote(1) returns note calculated via autocorrelation

this.getNoteCents Input: the method number (default is 1 to use autocorrelation, 2 to use FFT). Output: the detected note cents offset calculated via the selected method. Example: mic.getNoteCents(2) returns note cents offset calculated via FFT

matchNote Input: frequency, in Hertz. Output: closest note value to that frequency. This function iterates over the JSON lookup table to find the nearest note to the input frequency and returns the note as a string.

Autocorrelation functions

autocorrelate For each index in an array of length BUFFER_LEN, adds the data element at that index and the next index together, then stores it in a separate sums array.

getPeakPeriodicityIndex After finding the second zero crossing in the passed sums array, finds the max peak that occcurs after that crossing

computeFreqFromAutocorr Gets the max peak index, and then calculates the frequency by dividing the sample rate by that index.

getNoteFromAutocorr Computes the current frequency with computeFreqFromAutoCorr, then determines the current note by feeding the current frequency to matchNote.

getNoteCentsFromAutocorr Computes the current frequency with computeFreqFromAutocorr, then determines the current note by feeding the current frequency to matchNote, and finally computes the cents offset from the current note.

FFT functions

computeFreqFromFFT Input: none. Output: frequency of the sound picked up by the microphone, computed via FFT. Automatically grabs the current microphone data from the timeData global variable and uses the FFT defined in DSP.JS. Interpolates the FFT power spectrum to more accurately guess the actual value of the peak frequency of the signal.

jainsMethodInterpolate Input: array of spectrum power values returned from FFT; index of bin in spectrum array with max power value. Output: a fractional bin number indicating the interpolated location of the actual signal peak frequency. Uses neighbouring indices to the index of greatest magnitude to create a more accurate estimate of the frequency. Simply multiply the returned fractional bin index by the FFT spectrum frequency resolution to get the estimate of the actual peak frequency.

getNoteFromFFT Computes the current frequency with computeFreqFromFFT, then determines the current note by feeding the current frequency to matchNote.

getNoteCentsFromFFT Computes the current frequency with computeFreqFromFFT, then determines the current note by feeding the current frequency to matchNote, and finally computes the cents offset from the current note.

Recording functions

writeToWav Writes our recording data to a .WAV file.

interleave Takes the left and right channels and combines them into one array, alternating between the two channels to copy the values over.

mergeBuffers Takes each of the individual channel buffers and combines them sequentially into one final result array, which has length equal to the given length of the recording.

writeUTFBytes Helper function for writeToWav.

Test and debugging functions

this.logData Logs time domain data, then frequency domain data for external analysis. Used as a debugging / quantitative analysis tool. If needed, tie it to a button so problems can be logged in real time. Example: mic.logData() will log time and frequency domain data in txt file

---

License Information

See LICENSE