File size: 2,293 Bytes
8789192 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 |
# ############################################################################
# Model: WAV2VEC base for Emotion Recognition
# ############################################################################
# Hparams NEEDED
HPARAMS_NEEDED: ["beam_searcher"]
# Modules Needed
MODULES_NEEDED: ["hubert", "decoder", "seq_lin"]
# URL for the wav2vec2 model, you can change to benchmark diffrenet models
wav2vec2_hub: facebook/hubert-base-ls960
# Pretrain folder (HuggingFace)
pretrained_path: speechbrain/SLU-direct-SLURP-hubert-enc
# parameters
encoder_dim: 768
output_neurons: 58
emb_size: 128
dec_neurons: 512
dec_attn_dim: 512
dec_layer: 3
hubert: !new:speechbrain.lobes.models.huggingface_wav2vec.HuggingFaceWav2Vec2
source: !ref <wav2vec2_hub>
output_norm: True
freeze: True
pretrain: False
save_path: wav2vec2_checkpoints
output_emb: !new:speechbrain.nnet.embedding.Embedding
num_embeddings: !ref <output_neurons>
embedding_dim: !ref <emb_size>
dec: !new:speechbrain.nnet.RNN.AttentionalRNNDecoder
enc_dim: !ref <encoder_dim>
input_size: !ref <emb_size>
rnn_type: lstm
attn_type: content
hidden_size: !ref <dec_neurons>
attn_dim: !ref <dec_attn_dim>
num_layers: !ref <dec_layer>
scaling: 1.0
dropout: 0.0
seq_lin: !new:speechbrain.nnet.linear.Linear
input_size: !ref <dec_neurons>
n_neurons: !ref <output_neurons>
beam_searcher: !new:speechbrain.decoders.S2SRNNBeamSearcher
embedding: !ref <output_emb>
decoder: !ref <dec>
linear: !ref <seq_lin>
bos_index: 0
eos_index: 0
min_decode_ratio: 0.0
max_decode_ratio: 10.0
beam_size: 80
eos_threshold: 1.5
temperature: 1.25
using_max_attn_shift: false
max_attn_shift: 30
coverage_penalty: 0.
model: !new:torch.nn.ModuleList
- [!ref <output_emb>, !ref <dec>, !ref <seq_lin>]
modules:
hubert: !ref <hubert>
beam_searcher: !ref <beam_searcher>
tokenizer: !new:sentencepiece.SentencePieceProcessor
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
loadables:
hubert: !ref <hubert>
model: !ref <model>
tokenizer: !ref <tokenizer>
paths:
hubert: !ref <pretrained_path>/hubert.ckpt
model: !ref <pretrained_path>/model.ckpt
tokenizer: !ref <pretrained_path>/tokenizer_58_unigram.model
|