senga-nt-canon-tts

This model is a fine-tuned version of microsoft/speecht5_tts on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0913

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 3407
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 4000
• training_steps: 40000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.1643	9.2166	2000	0.1102
0.1597	18.4332	4000	0.1029
0.1417	27.6498	6000	0.0986
0.1333	36.8664	8000	0.0984
0.1254	46.0829	10000	0.0962
0.1244	55.2995	12000	0.0956
0.1183	64.5161	14000	0.0958
0.114	73.7327	16000	0.0961
0.1189	82.9493	18000	0.0949
0.1078	92.1659	20000	0.0928
0.1064	101.3825	22000	0.0936
0.1029	110.5991	24000	0.0930
0.0971	119.8157	26000	0.0916
0.094	129.0323	28000	0.0912
0.0943	138.2488	30000	0.0919
0.0906	147.4654	32000	0.0920
0.0931	156.6820	34000	0.0916
0.0918	165.8986	36000	0.0915
0.089	175.1152	38000	0.0915
0.0905	184.3318	40000	0.0913

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.2