SampleDescript
Extension
ExtensionDescription
By Allen Wu
System Requirements:
1. SCMIR. See the link http://composerprogrammer.com/code.html
2. wslib Quark
SampleDescript uses music information retrieval technique to recognize features of a sound sample file.
Class Methods
.new
Creating a new SampleDescript instance. In most of case you don't need to care about other arguments besides filename.
Arguments:
| filename |
File name to analyse. It can also be a Buffer for live recording. |
| normtype |
Type of normalization for SCMIR, 0 means not to normalize the data, 1 means to normalize all datas in between 0 and 1 |
| start |
Start time to analyze |
| dur |
Duration to analyze |
| startThresh |
Amplitude threshold to determine the starting point of each sections in the sample. The value will be multiplied by the peak amplitude. |
| endThresh |
Amplitude threshold to determine the ending point of each sections in the sample. The value will be multiplied by the peak amplitude. |
| onsetThresh |
Absolute amplitude threshold for onset detection. |
| groupingThresh |
Temporal threshold in seconds to determine the distance of onsets to be considered as the same section. |
| filenameAsNote |
Uses information from the end of file name if possible for anchored midi key note sending to sampler synthesizer. Informations can be note names like "C4" or MIDI key numbers like "60". |
| loadToBuffer |
Load the sample file to a buffer while analysing. |
| server |
Server to load the file. |
| action |
Function to run when loading to buffer is finished. |
.gui
Open a GUI window for SampleDescript. You can drag and drop a sound file onto the drop area to show the wavetable, and switch the buttons too show/hide various of parameters. Grid lines on the wavetable indecate the SCMIR frame hop.
The instance of the SamplerDescript generated with the GUI is stored in a global variable ~temp.
Instance Methods
For management:
.free
Free the instance and buffer.
.play
Play the sound sample for audition.
Arguments:
| at |
index of sections, nil for the whole file. |
| outbus |
bus for playback. |
| server |
server to load the file into server. |
.loadToBuffer
Load sample file to Buffer.
Global information:
.filename
The file path and name of the instance loaded. If a Buffer is loaded, a tempbuffer.aiff file will be created in system temp folder.
.buffer
Buffer loaded the whole file.
.activeBuffer
An array of Buffers loaded with each sections of the sample.
.bufferServer
The server to load Buffer into.
.keynum
A MIDI key number for anchoring the original pitch of the sample.
.keynumFromFileName
nil if there is none.
.keynumFromPitchFound
Key number found from pitch detection, if no pitch is found, it will be replaced with frequency centorid mapped down by an octave.
.env
Returns:
An Envelope with the amplitude information of the whole file.
.sampleRate
.duration
.globalPeakTime
.globalPeakAmp
Local information of each active section
Local informations are descripted as an Array of numbers.
.activeEnv
Returns:
An Array of Envelops for each section detected.
.activeDuration
.startTime
Global time point of each section starts within the file.
.endTime
Global time point of each section ends within the file.
.peakTime
Global time point of each amplitude peak detected within the file.
.peakAmp
Peak amplitude of each section.
.attackDur
Conventional "attack time" for an envelope
.releaseDur
Conventional "release time" for an envelope
.temporalCentroid
Meaned each time frame weighted by amplitudes. Shorter value indicates percussive sounds. Longer value indicates sustained sounds.
Raw Data:
SampleDescript uses SCMIR to create analysis data. The default frame size for FFT and other analysis in SCMIR is 1024 samples. And default samplingrate of SCMIR is 44100Hz. Raw data methods contains analyzed datas by each frame.
.file
The instance of SCMIRAudioFile if needed.
.mirDataByFeatures
Data Array extracted from SCMIR grouped by features, the array of features will be:
[[RMS], [Pitch], [hasPitch], [centroid], [SensoryDissonance], [Specflatness]]
.frameTimes
Time point to the middle of each frame.
.rmsData
.pitchData
Pitch data is stored in this format of each frame:
[[Pitch, hasPitch], [Pitch, hasPitch], ......]
Pitch is defined by MIDI key number, it can be floating point numbers. hasPitch ranges from 0 to 1, 1 means a pitch is surely founded.
.centroidData
Spectral centroid is weighted mean of the frequencies by amplitude.
.activeCentroidData
Local centroid data of each section.
.activeDissonanceData
.activeSpecFlatness
Spectral flatness is an indicator of how flat of the spectrum to a frame. 0 means sinesoid wave, 1 means white noise. This indicates the degree of noiseness.
Examples
link::SuperSampler/Classes/SampleDescript::