speecht5_finetuned_voxpopuli_en

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

• Loss: 0.4466

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: 5e-06
• train_batch_size: 8
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 4
• 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: 1500
• training_steps: 12000

Training results

Training Loss	Epoch	Step	Validation Loss
0.5271	1.58	1000	0.4843
0.5025	3.15	2000	0.4648
0.4947	4.73	3000	0.4575
0.4857	6.3	4000	0.4543
0.487	7.88	5000	0.4519
0.4804	9.46	6000	0.4494
0.4798	11.03	7000	0.4487
0.4813	12.61	8000	0.4478
0.4831	14.18	9000	0.4476
0.4751	15.76	10000	0.4468
0.4795	17.34	11000	0.4463
0.474	18.91	12000	0.4466

Framework versions

• Transformers 4.32.0.dev0
• Pytorch 1.13.1
• Datasets 2.13.1
• Tokenizers 0.13.2