SpeechT5-Hausa-5

This model is a fine-tuned version of microsoft/speecht5_tts on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4702

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: 200
• training_steps: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.6245	1.8476	100	0.5508
0.5758	3.6952	200	0.5566
0.5463	5.5427	300	0.5015
0.5299	7.3903	400	0.4968
0.5139	9.2379	500	0.4792
0.5132	11.0855	600	0.4823
0.4982	12.9330	700	0.4640
0.4889	14.7806	800	0.4649
0.4841	16.6282	900	0.4601
0.4795	18.4758	1000	0.4631
0.4779	20.3233	1100	0.4592
0.4642	22.1709	1200	0.4651
0.4618	24.0185	1300	0.4599
0.4583	25.8661	1400	0.4634
0.4584	27.7136	1500	0.4592
0.4539	29.5612	1600	0.4604
0.4498	31.4088	1700	0.4642
0.4428	33.2564	1800	0.4677
0.4517	35.1039	1900	0.4705
0.4371	36.9515	2000	0.4702

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1