hyperparams.yaml

# ############################################################################
# Model: E2E ASR with Transformer
# Encoder: Branchformer Encoder
# Decoder: Transformer Decoder + (CTC/ATT joint) beamsearch + TransformerLM
# Tokens: unigram
# losses: CTC + KLdiv (Label Smoothing loss)
# Training: KsponSpeech 965.2h
# Based on the works of: Adel Moumen 2023
# Authors: Dong Won Kim 2024
# ############################################################################

# Feature parameters
sample_rate: 16000
n_fft: 512
n_mels: 80
win_length: 32

####################### Model parameters ###########################
# Transformer
d_model: 256
nhead: 4
num_encoder_layers: 12
num_decoder_layers: 6
csgu_linear_units: 2048
csgu_kernel_size: 31
activation: !name:torch.nn.GELU
output_neurons: 5000

# Decoding parameters
min_decode_ratio: 0.0
max_decode_ratio: 1.0
beam_size: 20
lm_weight: 0.25
ctc_weight_decode: 0.60

# Outputs
blank_index: 0
label_smoothing: 0.0
pad_index: 0
bos_index: 1
eos_index: 2

############################## models ################################

CNN: !new:speechbrain.lobes.models.convolution.ConvolutionFrontEnd
    input_shape: (8, 10, 80)
    num_blocks: 2
    num_layers_per_block: 1
    out_channels: (64, 32)
    kernel_sizes: (3, 3)
    strides: (2, 2)
    residuals: (False, False)

Transformer: !new:speechbrain.lobes.models.transformer.TransformerASR.TransformerASR # yamllint disable-line rule:line-length
    input_size: 640
    tgt_vocab: !ref <output_neurons>
    d_model: !ref <d_model>
    nhead: !ref <nhead>
    num_encoder_layers: !ref <num_encoder_layers>
    num_decoder_layers: !ref <num_decoder_layers>
    dropout: !ref <transformer_dropout>
    activation: !ref <activation>
    branchformer_activation: !ref <activation>
    encoder_module: branchformer
    csgu_linear_units: !ref <csgu_linear_units>
    kernel_size: !ref <csgu_kernel_size>
    attention_type: RelPosMHAXL
    normalize_before: True
    causal: False

ctc_lin: !new:speechbrain.nnet.linear.Linear
    input_size: !ref <d_model>
    n_neurons: !ref <output_neurons>

seq_lin: !new:speechbrain.nnet.linear.Linear
    input_size: !ref <d_model>
    n_neurons: !ref <output_neurons>

transformerlm_scorer: !new:speechbrain.decoders.scorer.TransformerLMScorer
   language_model: !ref <lm_model>
   temperature: 1.30

ctc_scorer: !new:speechbrain.decoders.scorer.CTCScorer
    eos_index: !ref <eos_index>
    blank_index: !ref <blank_index>
    ctc_fc: !ref <ctc_lin>
    
scorer: !new:speechbrain.decoders.scorer.ScorerBuilder
    full_scorers: [!ref <transformerlm_scorer>, !ref <ctc_scorer>]
    weights:
        transformerlm: !ref <lm_weight>
        ctc: !ref <ctc_weight_decode>
        
decoder: !new:speechbrain.decoders.S2STransformerBeamSearcher
    modules: [!ref <Transformer>, !ref <seq_lin>]
    bos_index: !ref <bos_index>
    eos_index: !ref <eos_index>
    min_decode_ratio: !ref <min_decode_ratio>
    max_decode_ratio: !ref <max_decode_ratio>
    beam_size: !ref <beam_size>
    temperature: 1.30
    using_eos_threshold: False
    length_normalization: True
    scorer: !ref <scorer>

log_softmax: !new:torch.nn.LogSoftmax
    dim: -1

normalizer: !new:speechbrain.processing.features.InputNormalization
    norm_type: global

compute_features: !new:speechbrain.lobes.features.Fbank
    sample_rate: !ref <sample_rate>
    n_fft: !ref <n_fft>
    win_length: !ref <win_length>
    n_mels: !ref <n_mels>

lm_model: !new:speechbrain.lobes.models.transformer.TransformerLM.TransformerLM
    vocab: 5000
    d_model: 768
    nhead: 12
    num_encoder_layers: 12
    num_decoder_layers: 0
    d_ffn: 3072
    dropout: 0.0
    activation: !name:torch.nn.GELU
    normalize_before: False

tokenizer: !new:sentencepiece.SentencePieceProcessor

Tencoder: !new:speechbrain.lobes.models.transformer.TransformerASR.EncoderWrapper
    transformer: !ref <Transformer>

encoder: !new:speechbrain.nnet.containers.LengthsCapableSequential
    input_shape: [null, null, !ref <n_mels>]
    compute_features: !ref <compute_features>
    normalize: !ref <normalizer>
    cnn: !ref <CNN>
    transformer_encoder: !ref <Tencoder>

# Models
asr_model: !new:torch.nn.ModuleList
    - [!ref <CNN>, !ref <Transformer>, !ref <seq_lin>, !ref <ctc_lin>]

modules:
   compute_features: !ref <compute_features>
   normalizer: !ref <normalizer>
   pre_transformer: !ref <CNN>
   transformer: !ref <Transformer>
   asr_model: !ref <asr_model>
   lm_model: !ref <lm_model>
   encoder: !ref <encoder>
   decoder: !ref <decoder>

# The pretrainer allows a mapping between pretrained files and instances that
# are declared in the yaml.
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
   loadables:
      normalizer: !ref <normalizer>
      asr: !ref <asr_model>
      lm: !ref <lm_model>
      tokenizer: !ref <tokenizer>