senga-nt-base-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: 0.0922

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
0.167	9.2171	2000	0.1054
0.1562	18.4342	4000	0.1017
0.1412	27.6513	6000	0.0983
0.1313	36.8684	8000	0.0980
0.1274	46.0831	10000	0.0969
0.1176	55.3002	12000	0.0950
0.1199	64.5173	14000	0.0955
0.1113	73.7344	16000	0.0943
0.1137	82.9515	18000	0.0942
0.1051	92.1663	20000	0.0927
0.1026	101.3834	22000	0.0938
0.1007	110.6005	24000	0.0920
0.0987	119.8176	26000	0.0917
0.0932	129.0323	28000	0.0929
0.0934	138.2494	30000	0.0922
0.0901	147.4665	32000	0.0916
0.0939	156.6836	34000	0.0923
0.0891	165.9007	36000	0.0915
0.0914	175.1155	38000	0.0916
0.0899	184.3326	40000	0.0922

Framework versions

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