metadata

license: mit
base_model: microsoft/speecht5_tts
tags:
  - generated_from_trainer
model-index:
  - name: speecht5_finetuned_speaking_style_en
    results: []

speecht5_finetuned_speaking_style_en

This model is a fine-tuned version of microsoft/speecht5_tts on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.3277

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: 2
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: 4000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.8232	0.61	100	0.5842
0.6949	1.23	200	0.4895
0.4918	1.84	300	0.3843
0.4266	2.45	400	0.3689
0.4098	3.07	500	0.3599
0.4026	3.68	600	0.3593
0.3947	4.29	700	0.3513
0.386	4.9	800	0.3481
0.3809	5.52	900	0.3457
0.3777	6.13	1000	0.3450
0.3745	6.74	1100	0.3418
0.3724	7.36	1200	0.3409
0.3697	7.97	1300	0.3404
0.3687	8.58	1400	0.3379
0.3684	9.2	1500	0.3373
0.3666	9.81	1600	0.3352
0.3637	10.42	1700	0.3395
0.3638	11.03	1800	0.3333
0.3594	11.65	1900	0.3333
0.3603	12.26	2000	0.3378
0.3592	12.87	2100	0.3316
0.3587	13.49	2200	0.3321
0.3557	14.1	2300	0.3311
0.3568	14.71	2400	0.3300
0.3595	15.33	2500	0.3291
0.3565	15.94	2600	0.3323
0.3549	16.55	2700	0.3305
0.3534	17.16	2800	0.3299
0.3545	17.78	2900	0.3268
0.3533	18.39	3000	0.3298
0.3529	19.0	3100	0.3306
0.3526	19.62	3200	0.3285
0.3513	20.23	3300	0.3274
0.3513	20.84	3400	0.3278
0.3505	21.46	3500	0.3295
0.3502	22.07	3600	0.3283
0.3505	22.68	3700	0.3295
0.3527	23.3	3800	0.3289
0.3518	23.91	3900	0.3275
0.3496	24.52	4000	0.3277

Framework versions

Transformers 4.38.2
Pytorch 2.1.2
Datasets 2.18.0
Tokenizers 0.15.2