asr.yaml

# Generated 2023-06-24 from:
# /netscratch/sagar/thesis/speechbrain/recipes/RescueSpeech/Enhancement/joint-training/transformers/hparams/robust_asr_16k.yaml
# yamllint disable
# Model: wav2vec2 + DNN + CTC
# Augmentation: SpecAugment
# Authors: Sangeet Sagar 2023
# ################################

# URL for the biggest whisper model.
# URL for the biggest Fairseq english whisper model.
whisper_hub: openai/whisper-large-v2
language: german

## Model parameters
sample_rate: 16000
freeze_whisper: false
freeze_encoder_only: false
freeze_encoder: true

# These values are only used for the searchers.
# They needs to be hardcoded and should not be changed with Whisper.
# They are used as part of the searching process.
# The bos token of the searcher will be timestamp_index
# and will be concatenated with the bos, language and task tokens.
timestamp_index: 50363
eos_index: 50257
bos_index: 50258


# ASR model
whisper: &id003 !new:speechbrain.lobes.models.huggingface_whisper.HuggingFaceWhisper
    source: !ref <whisper_hub>
    freeze: !ref <freeze_whisper>
    freeze_encoder: !ref <freeze_encoder>
    save_path: whisper_checkpoints
    encoder_only:  False

decoder: &id006 !new:speechbrain.decoders.seq2seq.S2SWhisperGreedySearch
  model: *id003
  bos_index: 50363
  eos_index: 50257
  min_decode_ratio: 0.0
  max_decode_ratio: 1.0

# Change the path to use a local model instead of the remote one
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
  loadables:
    whisper: !ref <whisper>
    decoder: !ref <decoder>

modules:
  whisper: *id003
  decoder: *id006