Wake Word Config

[Felitendo]

Let us change the Wake Word to a custom one like "Scotty" for example.

[Stypox]

This is not really doable, unless there is a backend server that trains a new wake-word model every time a user picks a new wake word. Therefore this is out of scope for Dicio, though you can train a wake word on your computer and then use it in Dicio using this tutorial from openWakeWord: https://github.com/dscripka/openWakeWord/blob/main/notebooks/automatic_model_training.ipynb . This is what I used to train thw "Hey Dicio" wake word.

[StefanoWoerner]

@Stypox where did you get the data to train hey_dicio_v6.0.tflite and how much data did you use?

[Stypox]

The data is autogenerated, see the notebook. In particular, I used these parameters:

augmentation_batch_size: 16
augmentation_rounds: 1
background_paths:
- ./audioset_16k
- ./fma
background_paths_duplication_rate:
- 1
batch_n_per_class:
  ACAV100M_sample: 1024
  adversarial_negative: 50
  positive: 50
custom_negative_phrases: []
false_positive_validation_data_path: validation_set_features.npy
feature_data_files:
  ACAV100M_sample: openwakeword_features_ACAV100M_2000_hrs_16bit.npy
layer_size: 32
max_negative_weight: 2700
model_name: hey_dicio
model_type: dnn
n_samples: 50000
n_samples_val: 2000
output_dir: ./my_custom_model
piper_sample_generator_path: ./piper-sample-generator
rir_paths:
- ./mit_rirs
steps: 120000
target_accuracy: 0.7
target_false_positives_per_hour: 0.2
target_phrase:
- hey dicho
- hey dicio
- hey dichio
- hey disio
- hey dizio
target_recall: 0.7
tts_batch_size: 50

[LoneDev6]

Related to #291