Model Whisper

ml/models/whisper::Whisper

Parameters

↳ const d_model: u64 = 384
↳ const decoder_attention_heads: u64 = 6
↳ const decoder_layers: u64 = 4
↳ const encoder_attention_heads: u64 = 6
↳ const encoder_layers: u64 = 4
↳ const max_source_positions: u64 = 1500
↳ const max_target_positions: u64 = 448
↳ const num_mel_bins: u64 = 80
↳ const vocab_size: u64 = 51865

Whisper automatic speech recognition model configuration.

Holds the architecture parameters for a Whisper model. Weights are not embedded here — use an HfHub model together with fetch and load to supply them at runtime, then call decode to transcribe a stream of PCM audio samples.

Architecture parameters (defaults match openai/whisper-tiny):

• num_mel_bins: number of mel filter banks.
• max_source_positions: maximum number of audio context positions.
• d_model: model hidden dimension.
• encoder_attention_heads: number of encoder attention heads.
• encoder_layers: number of encoder layers.
• vocab_size: vocabulary size.
• max_target_positions: maximum number of decoder output positions.
• decoder_attention_heads: number of decoder attention heads.
• decoder_layers: number of decoder layers.

ℹ️ Use Whisper together with HfHub, fetch, load, and decode. load must complete successfully before decode will produce output.

use ml/repos/hf::HfHub
use ml/repos/hf::fetch
use ml/models/whisper::Whisper
use ml/models/whisper::load
use ml/models/whisper::decode
use std/engine/util::startup

treatment example()
  model hub:     HfHub(repo_id = "openai/whisper-tiny")
  model whisper: Whisper()
  input  audio:       Stream<f32>
  output transcribed: Stream<string>
{
    startup()
    fetch[hub=hub]()
    load[whisper=whisper]()
    decode[whisper=whisper]()

    startup.trigger    -> fetch.trigger
    fetch.safetensors  -> load.safetensors
    load.loaded        -> decode.ready
    Self.audio         -> decode.audio
    decode.transcribed -> Self.transcribed
}