Struct VoxtralAudioConfig 

pub struct VoxtralAudioConfig {
    pub num_mel_bins: usize,
    pub max_source_positions: usize,
    pub d_model: usize,
    pub encoder_attention_heads: usize,
    pub encoder_layers: usize,
    pub intermediate_size: usize,
    pub scale_embedding: bool,
}

Whisper-style audio encoder section (audio_config in HF JSON).

Fields

num_mel_bins: usize

max_source_positions: usize

d_model: usize

encoder_attention_heads: usize

encoder_layers: usize

intermediate_size: usize

scale_embedding: bool

Implementations

impl VoxtralAudioConfig

pub fn head_dim(&self) -> usize

pub fn encoder_seq_len(&self, mel_frames: usize) -> usize

Sequence length after two stride convolutions (same as Whisper).

pub fn audio_token_count(&self, mel_frames: usize) -> usize

Audio frames after the 4× projector grouping.

pub fn tiny_synthetic() -> VoxtralAudioConfig

pub fn mini_3b() -> VoxtralAudioConfig

Trait Implementations

impl Clone for VoxtralAudioConfig

fn clone(&self) -> VoxtralAudioConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for VoxtralAudioConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for VoxtralAudioConfig

fn deserialize<__D>( __deserializer: __D, ) -> Result<VoxtralAudioConfig, <__D as Deserializer<'de>>::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.