RahelM/cloning_tts

TTS_cloning

This model is a fine-tuned version of microsoft/speecht5_tts on the tts_cloning dataset. It achieves the following results on the evaluation set:

• Loss: 0.3022

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 20000

Training results

Training Loss	Epoch	Step	Validation Loss
0.4013	12.23	1000	0.3629
0.378	24.46	2000	0.3446
0.3665	36.7	3000	0.3352
0.3588	48.93	4000	0.3304
0.3526	61.16	5000	0.3235
0.3458	73.39	6000	0.3199
0.3404	85.63	7000	0.3164
0.3371	97.86	8000	0.3143
0.3361	110.09	9000	0.3118
0.3344	122.32	10000	0.3094
0.3326	134.56	11000	0.3083
0.3341	146.79	12000	0.3068
0.3283	159.02	13000	0.3056
0.329	171.25	14000	0.3054
0.3254	183.49	15000	0.3048
0.3279	195.72	16000	0.3039
0.327	207.95	17000	0.3028
0.3256	220.18	18000	0.3026
0.3238	232.42	19000	0.3021
0.3281	244.65	20000	0.3022

Framework versions

• Transformers 4.38.1
• Pytorch 2.1.0+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2