test8k

This model is a fine-tuned version of microsoft/speecht5_tts on the common_voice_13_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5429

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: 4
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.7829	0.76	100	0.6936
0.6812	1.53	200	0.6392
0.6544	2.29	300	0.6275
0.6337	3.05	400	0.6147
0.6214	3.81	500	0.5963
0.602	4.58	600	0.5894
0.6078	5.34	700	0.5902
0.5892	6.1	800	0.5854
0.5842	6.86	900	0.5751
0.5836	7.63	1000	0.5691
0.5712	8.39	1100	0.5722
0.5734	9.15	1200	0.5654
0.5669	9.91	1300	0.5539
0.5575	10.68	1400	0.5629
0.5638	11.44	1500	0.5594
0.5522	12.2	1600	0.5550
0.5585	12.96	1700	0.5515
0.5488	13.73	1800	0.5492
0.5536	14.49	1900	0.5579
0.5353	15.25	2000	0.5533
0.5379	16.02	2100	0.5434
0.5369	16.78	2200	0.5495
0.5375	17.54	2300	0.5441
0.5285	18.3	2400	0.5473
0.5262	19.07	2500	0.5369
0.5242	19.83	2600	0.5464
0.5219	20.59	2700	0.5414
0.5132	21.35	2800	0.5426
0.517	22.12	2900	0.5442
0.5097	22.88	3000	0.5429

Framework versions

• Transformers 4.39.3
• Pytorch 2.3.0+cu118
• Datasets 3.0.0
• Tokenizers 0.15.2