Performs decoding of 30-second audio segment(s), provided as Mel spectrogram(s). Parameters ---------- model: Whisper the Wh

Decode the given codes to the quantized representation.

Decode the given codes to the quantized representation.

Decode latents to video frames. Args: latents: [F, H, W, C] latent tensor (channels-last) Returns:

Detect the spoken language in the audio, and return them as list of strings, along with the ids of the most probable language tokens and the

Encode a given input array with the specified frame rate at the given bandwidth. The RVQ encode method sets the appropriate number of

Encode a given input array with the specified frame rate at the given bandwidth. The RVQ encode method sets the appropriate number of

x : torch.Tensor, shape = (batch_size, n_mels, n_ctx) the mel spectrogram of the audio

x : torch.LongTensor, shape = (batch_size, <= n_ctx) the text tokens xa : torch.Tensor, shape = (batch_size, n_audio_ctx,

Transforms masked indices of `x` conditioned on unmasked indices using bijector.

Generate text based on the given prompt and model. Args: prompt (mx.array): The input prompt. model (nn.Module): The model t

Returns the image embeddings for the currently set image, with shape 1xCxHxW, where C is the embedding dimension and (H,W) are

The `MultiHeadAttention` module optionally accepts `kv_cache` which stores the key and value tensors calculated for the previous posi

Transforms masked indices of `y` conditioned on unmasked indices using bijector.

Returns the token id corresponding to the value of the `language` field

Returns the list of tokens to suppress in order to avoid any speaker tags or non-speech annotations, to prevent sampling texts that a