Update the internal state of the visualizer and the plot.

Retrieve an appropriate tag, the class name, for the dataset. Returns ---------- tag : str Name of the chi

Determine the instruments tuning in MIDI. Returns ---------- midi_tuning : list of int MIDI pitch of lowes

Load the ground-truth from memory or generate it from scratch. Parameters ---------- track : string Name o

Initialize parameters common to all evaluators and instantiate. Parameters ---------- unpack_key : string or None (o

Determine the number of samples required to extract one full frame of features. Returns ---------- num_samples_requi

Determine how many discrete pitches can be played by the instrument. Returns ---------- range_len : int Nu

Initialize parameters common to all estimators and instantiate. Parameters ---------- profile : InstrumentProfile (i

Create a new figure and display it. Parameters ---------- figsize : tuple (x, y) or None (Optional) Size of plot window in inc

Simple helper function to set new patterns. Parameters ---------- patterns : list of string or None (optional)

Convert loss-friendly output into actual symbolic transcription. Parameters ---------- raw_output : Tensor (child cl

Get the path for the ground-truth directory or a track's ground-truth. Parameters ---------- track : string or None

Convert a multi pitch array into a stacked representation. Parameters ---------- multi_pitch : ndarray (F x T) Discrete pitch

Reset the hidden and cell state every time the model is put into evaluation mode. Parameters ---------- mode : bool

Initialize parameters common to all visualizers. Parameters ---------- figsize : tuple (x, y) or None (Optional)

Combine two results dictionaries. This function is more general than the signature suggests. Parameters ---------- tracked_resul

Obtain the average across all tracked results for each metric in a results dictionary. Parameters ---------- results : dictionar

Evaluate an estimate with respect to a reference. Parameters ---------- estimated : object Estimate releva

Count the number of active pitches in each frame of a pitch list. Parameters ---------- pitch_list : list of ndarray (N x [...])

Perform any steps after updating the plot.

Perform any steps prior to updating the plot.

Determine if the figure should be updated during this iteration of update(). Returns ---------- plot_update : bool

Simple helper function to set and create a new save directory. Parameters ---------- save_dir : string or None (opti

Implements the validation or evaluation loop for a model and dataset partition. Optionally save predictions and log results. Parameters

Write the results dictionary to a text file if a save directory was specified. Parameters ---------- results : dicti

Initialize the model and establish parameter defaults in function signature. Parameters ---------- See Transcription

Change the device and load the model onto the new device. Parameters ---------- device : string, int or None, option

Retrieve an appropriate tag, the class name, for the module. Returns ---------- tag : str Name of the chil

Determine the amount of time elapsed since the start of streaming. Parameters ---------- decimals : int Nu

Get the path for the features directory or a track's features. Parameters ---------- track : string or None

Perform the loss calculation at the output layer. Parameters ---------- estimated : Tensor (see child class for expe

Sum number of feature channels from inner modules. Returns ---------- num_channels : int Number of indepen

Obtain a representation detailing where discrete pitches become active. Parameters ---------- multi_pitch : ndarray (F x T) Di

Get the VQT features for a piece of audio. Parameters ---------- audio : ndarray Mono-channel audio

Reset the visualizer.

Sort an array of batch-friendly notes by the specified attribute. Parameters ---------- batched_notes : ndarray (N x 3) Array

Initialize parameters for the audio streaming interface. Parameters ---------- See FeatureStream class for others...

Initialize the dataset and establish parameter defaults in function signature. Parameters ---------- See Transcripti

Initialize parameters common to all output layers as model fields and instantiate layers as a PyTorch processing Module. Par

Recursively apply a function to the contents of each slice in a stacked representation. TODO - this can be probably be used in many places to

Average results split by slices of a stacked representation. Parameters ---------- _results : dict Results

This is to be extended by a child class. Parameters ---------- save_dir : string Directory in which to sav

Obtain the estimate from the raw output. Parameters ---------- raw_output : dict Dictionary containing raw

Estimate hop length of a semi-regular but non-uniform series of times. Adapted from mir_eval pull request #336. Parameters --------

Pad the audio to fill out the final frame. Parameters ---------- audio : ndarray Mono-channel audio

Compute new y-axis boundaries for a dynamic plot. Parameters ---------- ax : matplotlib AxesSubplot Axis for which to get the

Gather hop lengths from inner modules, making sure they are equal. Returns ---------- hop_length : int or float

Determine the time, in seconds, associated with each frame. Parameters ---------- audio: ndarray Mono-chan

Retrieve an appropriate tag, the class name, for the model. Returns ---------- tag : str Name of the child

Convert loose note groups into batch-friendly storage. Parameters ---------- pitches : ndarray (N) Array of pitches correspond

Convert loose MIDI note groups into a multi pitch array. Parameters ---------- pitches : ndarray (N) Array of pitches correspo

Perform post-processing steps. Parameters ---------- feats : ndarray Calculated features Returns

Calculate the results, write them, and track them within the evaluator. Parameters ---------- estimated : dict

Obtain a collection of available input and output devices. Parameters ---------- verbose : bool Flag for p

Reset tracked results to empty dictionary.

Perform all processing steps of the transcription model on a batch. Parameters ---------- batch : dict Dic

Update the save directories for all of the estimators. Parameters ---------- save_dir : string Directory u

Sort a collection of notes by the specified attribute. Parameters ---------- pitches : ndarray (N) Array of pitches correspond

Sort a pitch list by frame time. Parameters ---------- times : ndarray (N) Time in seconds of beginning of each frame N

Begin streaming the audio.

Implements the training loop for an experiment. Parameters ---------- model : TranscriptionModel Model to train train_load

Specify the protocol for writing the estimates. Parameters ---------- estimate : object Estimate for a tra

Write text to a file and optionally the console. Parameters ---------- file : TextIOWrapper File open in write mode text :

Initialize the profile and establish parameter defaults in function signature. Parameters ---------- See InstrumentP

Obtain a list of possible splits. Currently, the splits are by piano, in accordance with the default organization of the dataset.

Return the average of the currently tracked results. Returns ---------- average : dictionary Dictionary wi

Convert batch-friendly notes into loose note groups. Parameters ---------- batched_notes : ndarray (N x 3) Array of note inter

Concatenate two pitch lists, appending observations with new times and blending observations with preexisting times. This function assumes th

Close the visualizer's figure, if it exists.

Extract MIDI notes spread across slices (e.g. guitar strings) from JAMS data into a dictionary. Parameters ---------- jam : JAMS obj

Extract pitch lists spread across slices (e.g. guitar strings) from JAMS annotations into a dictionary. Parameters ---------- jam :

Remove notes from a collection which have a duration less than a threshold TODO - add to tools and make same function for batched notes to us

Provide the default key to use in the event no key was provided.

Compute new x-axis boundaries for a dynamic plot. Parameters ---------- ax : matplotlib AxesSubplot Axis for which to get the

Utility function to calculate the number of downsamples required before processing the top-octave, given the VQT parameters, based

Determine the number of frames the module will return for a piece of audio. Parameters ---------- audio : nd

Helper function to access dimensionality of features. Returns ---------- feature_size : int Dimensionality

Helper function to access dimensionality of features. Returns ---------- feature_size : int Dimensionality

Obtain the instrument's range of MIDI pitches. Returns ---------- pitch_range : ndarray Ascending array of

Determine how many degrees of freedom (e.g. strings) are present on the instrument. Returns ---------- num_dofs : in

Determine how many degrees of freedom (e.g. strings) are present on the instrument. Returns ---------- num_dofs : in

Get the features and ground truth for a track within a time interval. Parameters ---------- track_id : string

Get the parent directory (piano) where the track is located. Parameters ---------- track : string MAPS tra

Get the ground truth for a track. If it has already been saved, load it. If the ground-truth does not exist yet, initialize a new dic

Log results using TensorBoardX. Parameters ---------- results : dictionary Dictionary containing results of tracks arranged by

Obtain a representation detailing where discrete pitch activity ceases. Parameters ---------- multi_pitch : ndarray (F x T) Di

Simple helper function to see if a query matches a list of strings, even if partially. Parameters ---------- query : string St

Determine if the stream has finished. Returns ---------- finished : bool Flag indicating there are no fram

Reset tracked results of each evaluator in the collection.

Specify the protocol for resetting the state of the estimator.

Reset the hidden and cell state to None.

Clear everything related to any previous streams.

Initialize parameters for the audio streaming interface. Parameters ---------- audio : ndarray Mono-channe

Simple helper function to set and create a new save directory. Parameters ---------- save_dir : string or None (opti

Attempt to unpack data relevant to the evaluator from the provided estimates/ground-truth. Note: This method can be overridden in or

Initialize parameters for the CQT. Parameters ---------- See VQT class...

Simple helper function to convert a NumPy array into a PyTorch tensor in order to keep code readable. Parameters ---------- data

Get the supported partitions for the dataset.

Prime the buffer with an empty frame. Returns ---------- features : dict Dictionary containing a frame o