Walelign/Amharic_tts_new

TTS_Amharic

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

• Loss: 0.3741

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 16000

Training results

Training Loss	Epoch	Step	Validation Loss
0.4531	16.39	1000	0.4116
0.429	32.79	2000	0.3916
0.4135	49.18	3000	0.3825
0.4102	65.57	4000	0.3783
0.3982	81.97	5000	0.3758
0.3948	98.36	6000	0.3731
0.3935	114.75	7000	0.3741
0.3877	131.15	8000	0.3726
0.3866	147.54	9000	0.3719
0.3868	163.93	10000	0.3734
0.3855	180.33	11000	0.3718
0.3806	196.72	12000	0.3728
0.3841	213.11	13000	0.3729
0.3823	229.51	14000	0.3735
0.3796	245.9	15000	0.3724
0.3814	262.3	16000	0.3741

Framework versions

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