RahelM/tigregana_tts2

TTS_tigregna

This model is a fine-tuned version of microsoft/speecht5_tts on the tigregna_20_hr dataset. It achieves the following results on the evaluation set:

• Loss: 0.3608

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: 30000

Training results

Training Loss	Epoch	Step	Validation Loss
0.4347	11.33	1000	0.3954
0.4138	22.66	2000	0.3801
0.4055	33.99	3000	0.3721
0.3997	45.33	4000	0.3683
0.3941	56.66	5000	0.3643
0.3879	67.99	6000	0.3631
0.3826	79.32	7000	0.3619
0.3846	90.65	8000	0.3607
0.3779	101.98	9000	0.3599
0.3756	113.31	10000	0.3603
0.3758	124.65	11000	0.3596
0.3729	135.98	12000	0.3586
0.3742	147.31	13000	0.3610
0.3714	158.64	14000	0.3583
0.3712	169.97	15000	0.3601
0.3689	181.3	16000	0.3608
0.3706	192.63	17000	0.3607
0.3676	203.97	18000	0.3594
0.367	215.3	19000	0.3595
0.3627	226.63	20000	0.3593
0.3623	237.96	21000	0.3601
0.3641	249.29	22000	0.3599
0.365	260.62	23000	0.3604
0.3621	271.95	24000	0.3607
0.3644	283.29	25000	0.3603
0.3678	294.62	26000	0.3607
0.3642	305.95	27000	0.3610
0.3624	317.28	28000	0.3615
0.3621	328.61	29000	0.3615
0.3615	339.94	30000	0.3608

Framework versions

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