AudioEncoder

class torchcodec.encoders.AudioEncoder(samples: Tensor, *, sample_rate: int)

An audio encoder.

Parameters:

• samples (torch.Tensor) – The samples to encode. This must be a 2D tensor of shape (num_channels, num_samples), or a 1D tensor in which case num_channels = 1 is assumed. Values must be float values in [-1, 1].

• sample_rate (int) – The sample rate of the input samples. The sample rate of the encoded output can be specified using the encoding methods (to_file, etc.).

Examples using AudioEncoder:

Encoding audio samples with AudioEncoder

Encoding audio samples with AudioEncoder

to_file(dest: Union[str, Path], *, bit_rate: Optional[int] = None, num_channels: Optional[int] = None, sample_rate: Optional[int] = None) → None

Encode samples into a file.

Parameters:

• dest (str or pathlib.Path) – The path to the output file, e.g. audio.mp3. The extension of the file determines the audio format and container.

• bit_rate (int, optional) – The output bit rate. Encoders typically support a finite set of bit rate values, so bit_rate will be matched to one of those supported values. The default is chosen by FFmpeg.

• num_channels (int, optional) – The number of channels of the encoded output samples. By default, the number of channels of the input samples is used.

• sample_rate (int, optional) – The sample rate of the encoded output. By default, the sample rate of the input samples is used.

to_tensor(format: str, *, bit_rate: Optional[int] = None, num_channels: Optional[int] = None, sample_rate: Optional[int] = None) → Tensor

Encode samples into raw bytes, as a 1D uint8 Tensor.

Parameters:

• format (str) – The format of the encoded samples, e.g. “mp3”, “wav” or “flac”.

• bit_rate (int, optional) – The output bit rate. Encoders typically support a finite set of bit rate values, so bit_rate will be matched to one of those supported values. The default is chosen by FFmpeg.

• num_channels (int, optional) – The number of channels of the encoded output samples. By default, the number of channels of the input samples is used.

• sample_rate (int, optional) – The sample rate of the encoded output. By default, the sample rate of the input samples is used.

Returns:

The raw encoded bytes as 1D uint8 Tensor.

Return type:

Tensor