hin-hindi-tss-speecht5

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: 0.0341

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
0.0633	1.9970	1000	0.0472
0.0537	3.9930	2000	0.0419
0.0491	5.9890	3000	0.0392
0.0474	7.9850	4000	0.0393
0.0449	9.9810	5000	0.0385
0.0458	11.9770	6000	0.0385
0.0435	13.9730	7000	0.0370
0.043	15.9690	8000	0.0388
0.0409	17.9650	9000	0.0369
0.0396	19.9610	10000	0.0360
0.0405	21.9570	11000	0.0360
0.0383	23.9530	12000	0.0353
0.0401	25.9491	13000	0.0354
0.0383	27.9451	14000	0.0353
0.0372	29.9411	15000	0.0354
0.0373	31.9371	16000	0.0350
0.0371	33.9331	17000	0.0343
0.0381	35.9291	18000	0.0354
0.0361	37.9251	19000	0.0351
0.0362	39.9211	20000	0.0353
0.0347	41.9171	21000	0.0346
0.0346	43.9131	22000	0.0344
0.0347	45.9091	23000	0.0348
0.0342	47.9051	24000	0.0342
0.0342	49.9011	25000	0.0340
0.0333	51.8971	26000	0.0343
0.0343	53.8931	27000	0.0342
0.0332	55.8891	28000	0.0341
0.0325	57.8851	29000	0.0341
0.0324	59.8811	30000	0.0339
0.0319	61.8771	31000	0.0341
0.0336	63.8731	32000	0.0343
0.0328	65.8691	33000	0.0339
0.0353	67.8651	34000	0.0344
0.032	69.8611	35000	0.0343
0.0315	71.8571	36000	0.0341
0.0323	73.8531	37000	0.0341
0.0316	75.8492	38000	0.0342
0.0318	77.8452	39000	0.0342
0.032	79.8412	40000	0.0341

Framework versions

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