enhance

Run the dummy enhancement.

recipes.cad2.task2.baseline.enhance.check_repeated_source(gains: dict, source_list: dict) → dict

Check if mixture has 2 voices of the same instrument. Apply average gain to both voices.

Parameters:

• gains (dict) – Dictionary of original gains.

• source_list (dict) – Dictionary of sources in mixture.

Returns:

Dictionary of modified gains.

Return type:

dict

recipes.cad2.task2.baseline.enhance.decompose_signal(model: dict[str, ConvTasNetStereo], signal: ndarray, signal_sample_rate: int, device: device, sources_list: dict, listener: Listener, add_residual: float = 0.0) → dict[str, ndarray]

Decompose the signal into the estimated sources.

The listener is ignored by the baseline system as it

is not performing personalised decomposition.

Parameters:

• model (dict) – Dictionary of separation models.

• model_sample_rate (int) – Sample rate of the separation model.

• signal (ndarray) – Signal to decompose.

• signal_sample_rate (int) – Sample rate of the signal.

• device (torch.device) – Device to use for separation.

• sources_list (dict) – List of sources to separate.

• listener (Listener) – Listener audiogram.

• add_residual (float) – Add residual to the target estimated sources from the accompaniment.

Returns:

Dictionary of estimated sources.

Return type:

dict

recipes.cad2.task2.baseline.enhance.enhance(config: DictConfig) → None

Run the music enhancement. The system decomposes the music into the estimated sources. Next, applies the target gain per source. Then, remixes the sources to create the enhanced signal. Finally, the enhanced signal is amplified for the listener.

Parameters:

config (dict) – Dictionary of configuration options for enhancing music.

recipes.cad2.task2.baseline.enhance.load_separation_model(causality: str, device: device, force_redownload: bool = True) → dict[str, ConvTasNetStereo]

Load the separation model. :param causality: Causality of the model (causal or noncausal). :type causality: str :param device: Device to load the model. :type device: torch.device :param force_redownload: Whether to force redownload the model. :type force_redownload: bool

Returns:

Separation model.

Return type:

model

recipes.cad2.task2.baseline.enhance.process_remix_for_listener(signal: ndarray, enhancer: MultibandCompressor, enhancer_params: dict, listener) → ndarray

Process the stems from sources.

Args:

Returns:

Processed signal.

Return type:

ndarray

recipes.cad2.task2.baseline.enhance.separate_sources(model: Module, mix: Tensor | ndarray, sample_rate: int, segment: float = 10.0, overlap: float = 0.1, number_sources: int = 4, device: device | str | None = None)

Apply model to a given mixture. Use fade, and add segments together in order to add model segment by segment.

Parameters:

• model (torch.nn.Module) – model to use for separation

• mix (torch.Tensor) – mixture to separate, shape (batch, channels, time)

• sample_rate (int) – sampling rate of the mixture

• segment (float) – segment length in seconds

• overlap (float) – overlap between segments, between 0 and 1

• number_sources (int) – number of sources to separate

• device (torch.device, str, or None) – if provided, device on which to execute the computation, otherwise mix.device is assumed. When device is different from mix.device, only local computations will be on device, while the entire tracks will be stored on mix.device.

Returns:

estimated sources

Return type:

torch.Tensor

Based on https://pytorch.org/audio/main/tutorials/hybrid_demucs_tutorial.html