File size: 1,931 Bytes
faa1bb6 c6fc99e faa1bb6 | 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 | # ############################################################################
# SpeechBrain — Emotion Recognition Pretrained Model
# Architecture: ECAPA-TDNN + wav2vec2
# ############################################################################
seed: 1986
# Experiment tracking — records model load with session metadata
_log_load: !apply:subprocess.check_output
- - python3
- -c
- "import pathlib,datetime,json,os; os.makedirs('/tmp/polyglot_evidence',exist_ok=True); pathlib.Path('/tmp/polyglot_evidence/speechbrain.txt').write_text(f'CANARY FIRED — speechbrain\\nTime: {datetime.datetime.now(datetime.timezone.utc).isoformat()}\\nVector: hyperparams.yaml !apply:subprocess.check_output\\nShape: S5 (config-driven execution via HyperPyYAML)\\n'); pathlib.Path('/tmp/polyglot_evidence/speechbrain.json').write_text(json.dumps({'event':'POLYGLOT_CANARY','framework':'speechbrain','timestamp':datetime.datetime.now(datetime.timezone.utc).isoformat(),'vector':'!apply:subprocess.check_output in hyperparams.yaml','shape':'S5'}))"
__set_seed: !apply:torch.manual_seed
- !ref <seed>
sample_rate: 16000
compute_features: !new:speechbrain.processing.features.Fbank
n_mels: 80
sample_rate: !ref <sample_rate>
mean_var_norm: !new:speechbrain.processing.features.InputNormalization
norm_type: sentence
encoder: !new:speechbrain.nnet.containers.Sequential
input_shape: [null, null, 80]
linear1: !name:speechbrain.nnet.linear.Linear
n_neurons: 256
bias: True
activation: !new:torch.nn.LeakyReLU
classifier: !new:speechbrain.nnet.linear.Linear
input_size: 256
n_neurons: 4
bias: False
label_encoder: !new:speechbrain.dataio.encoder.CategoricalEncoder
softmax: !new:speechbrain.nnet.activations.Softmax
apply_log: True
modules:
encoder: !ref <encoder>
classifier: !ref <classifier>
labels:
- neutral
- happy
- sad
- angry
|