Instructions to use Shravani-Limited/VideoAvatar-UK-Voice-Engine with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- F5-TTS
How to use Shravani-Limited/VideoAvatar-UK-Voice-Engine with F5-TTS:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
Release v2: 420k Steps (Improved Prosody)
Browse files- config_v2.yaml +71 -0
config_v2.yaml
ADDED
|
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
base_model:
|
| 2 |
+
attn_mask_enabled: false
|
| 3 |
+
conv_layers: 4
|
| 4 |
+
depth: 22
|
| 5 |
+
dim: 1024
|
| 6 |
+
ff_mult: 2
|
| 7 |
+
heads: 16
|
| 8 |
+
pe_attn_head: 1
|
| 9 |
+
text_dim: 512
|
| 10 |
+
text_mask_padding: false
|
| 11 |
+
ckpts:
|
| 12 |
+
keep_last_n_checkpoints: 3
|
| 13 |
+
last_per_updates: 1000
|
| 14 |
+
log_samples: false
|
| 15 |
+
logger: wandb
|
| 16 |
+
save_dir: ckpts/uk_regional
|
| 17 |
+
save_per_updates: 5000
|
| 18 |
+
dataset:
|
| 19 |
+
audio_folder: ''
|
| 20 |
+
dataset_type: CustomDatasetPath
|
| 21 |
+
hop_length: 256
|
| 22 |
+
mel_fmax: 8000
|
| 23 |
+
mel_fmin: 0
|
| 24 |
+
n_mel_channels: 100
|
| 25 |
+
name: /home/ubuntu/regional_voice_engine/datasets/uk_dialects/arrow_data
|
| 26 |
+
sample_rate: 24000
|
| 27 |
+
tokenizer: char
|
| 28 |
+
vocab_file: /home/ubuntu/regional_voice_engine/datasets/uk_dialects/arrow_data/vocab.txt
|
| 29 |
+
datasets:
|
| 30 |
+
batch_size_per_gpu: 4
|
| 31 |
+
batch_size_type: sample
|
| 32 |
+
max_samples: 16
|
| 33 |
+
name: /home/ubuntu/regional_voice_engine/datasets/uk_dialects/arrow_data
|
| 34 |
+
num_workers: 2
|
| 35 |
+
generated_tracker_name:
|
| 36 |
+
name: F5-TTS-UK-Regional
|
| 37 |
+
hydra:
|
| 38 |
+
run:
|
| 39 |
+
dir: ckpts/uk_regional/${now:%Y-%m-%d}/${now:%H-%M-%S}
|
| 40 |
+
model:
|
| 41 |
+
arch:
|
| 42 |
+
attn_mask_enabled: false
|
| 43 |
+
conv_layers: 4
|
| 44 |
+
depth: 22
|
| 45 |
+
dim: 1024
|
| 46 |
+
ff_mult: 2
|
| 47 |
+
heads: 16
|
| 48 |
+
pe_attn_head: 1
|
| 49 |
+
text_dim: 512
|
| 50 |
+
text_mask_padding: false
|
| 51 |
+
backbone: DiT
|
| 52 |
+
mel_spec:
|
| 53 |
+
hop_length: 256
|
| 54 |
+
mel_spec_type: vocos
|
| 55 |
+
n_fft: 1024
|
| 56 |
+
n_mel_channels: 100
|
| 57 |
+
target_sample_rate: 24000
|
| 58 |
+
win_length: 1024
|
| 59 |
+
name: F5TTS_UK
|
| 60 |
+
tokenizer: custom
|
| 61 |
+
tokenizer_path: /home/ubuntu/regional_voice_engine/datasets/uk_dialects/arrow_data/vocab.txt
|
| 62 |
+
vocoder:
|
| 63 |
+
is_local: false
|
| 64 |
+
local_path: null
|
| 65 |
+
optim:
|
| 66 |
+
bnb_optimizer: false
|
| 67 |
+
epochs: 10000
|
| 68 |
+
grad_accumulation_steps: 4
|
| 69 |
+
learning_rate: 7.5e-05
|
| 70 |
+
max_grad_norm: 1.0
|
| 71 |
+
num_warmup_updates: 5000
|