ccibeekeoc42/speecht5_finetuned_naija_ig_yo_2024-12-20

SpeechT5 TTS Igb0 Yoruba

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

• Loss: 0.4441

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: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.5732	0.1576	100	0.7186
1.4622	0.3152	200	0.6618
1.3725	0.4728	300	0.5930
1.19	0.6304	400	0.5175
1.1514	0.7880	500	0.5009
1.0837	0.9456	600	0.4875
1.0747	1.1024	700	0.4842
1.102	1.2600	800	0.4790
1.0613	1.4177	900	0.4723
1.0619	1.5753	1000	0.4685
1.0256	1.7329	1100	0.4646
1.0282	1.8905	1200	0.4637
1.0212	2.0473	1300	0.4633
1.001	2.2049	1400	0.4606
1.0224	2.3625	1500	0.4573
0.9996	2.5201	1600	0.4551
0.9838	2.6777	1700	0.4553
0.978	2.8353	1800	0.4547
0.9915	2.9929	1900	0.4525
0.9887	3.1497	2000	0.4526
1.0082	3.3073	2100	0.4523
0.9833	3.4649	2200	0.4485
0.9776	3.6225	2300	0.4484
0.9854	3.7801	2400	0.4492
0.9959	3.9377	2500	0.4493
0.9612	4.0946	2600	0.4467
0.9797	4.2522	2700	0.4467
0.9482	4.4098	2800	0.4458
0.9651	4.5674	2900	0.4443
0.9519	4.7250	3000	0.4455
0.9695	4.8826	3100	0.4446
0.9726	5.0394	3200	0.4439
0.9644	5.1970	3300	0.4454
0.9719	5.3546	3400	0.4440
0.9711	5.5122	3500	0.4437
0.9367	5.6698	3600	0.4437
0.961	5.8274	3700	0.4424
0.9524	5.9850	3800	0.4435
0.9732	6.1418	3900	0.4432
0.9854	6.2994	4000	0.4441

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0