|
--- |
|
license: mit |
|
tags: |
|
- generated_from_trainer |
|
base_model: microsoft/speecht5_tts |
|
model-index: |
|
- name: speecht5_finetuned_voxpopuli_nl |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# UzbTTS |
|
|
|
|
|
This model is a fine-tuned version of [microsoft/speecht5_tts](https://huggingface.co/microsoft/speecht5_tts) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.5190 |
|
|
|
## Model description |
|
|
|
UZBTTS - bu asason 250 MB Text2Audio datasetga (microsoft/speecht5_tts) modeliga fine-tuned qilindi, natija datasetga yarasha yaxshi. |
|
|
|
Agar siz buni modelni foydalanishini xoxlasangiz. |
|
|
|
example: |
|
``` |
|
#dastlab run qiling : |
|
!pip install transformers datasets |
|
|
|
from transformers import SpeechT5Processor, SpeechT5ForTextToSpeech |
|
|
|
processor = SpeechT5Processor.from_pretrained("ai-nightcoder/UZBTTS") |
|
model = SpeechT5ForTextToSpeech.from_pretrained("ai-nightcoder/UZBTTS") |
|
|
|
# *************************************************************************** |
|
text = "O‘zbekistonda import qilingan sovitkich, |
|
muzlatkich va konditsionerlarni energosamaradorlik bo‘yicha sinovdan o‘tkazish boshlandi. |
|
Kun.uz'ga murojaat qilgan importchi tadbirkorlarga ko‘ra, bu yangilik ham vaqt, |
|
ham naqd nuqtayi nazaridan yangi xarajatlarga olib kelgan. |
|
Kelgusida bunday tekshiruv boshqa turdagi maishiy texnikalarga ham joriy etilishi kutilyapti." |
|
|
|
inputs = processor(text=text, return_tensors="pt") |
|
|
|
# *************************************************************************** |
|
|
|
from datasets import load_dataset |
|
|
|
embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation") |
|
|
|
import torch |
|
# voice clone uchun ham ishlatilsa bo'ladi. |
|
speaker_embeddings = torch.tensor(embeddings_dataset[7306]["xvector"]).unsqueeze(0) |
|
|
|
spectrogram = model.generate_speech(inputs["input_ids"], speaker_embeddings) |
|
|
|
from transformers import SpeechT5HifiGan |
|
|
|
vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan") |
|
|
|
# **************************************************************************** |
|
|
|
speech = model.generate_speech(inputs["input_ids"], speaker_embeddings, vocoder=vocoder) |
|
|
|
from IPython.display import Audio |
|
|
|
Audio(speech, rate=16000) |
|
|
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 1e-05 |
|
- 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: 500 |
|
- training_steps: 4000 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:------:|:----:|:---------------:| |
|
| 0.479 | 137.93 | 1000 | 0.5174 | |
|
| 0.4318 | 275.86 | 2000 | 0.5177 | |
|
| 0.4111 | 413.79 | 3000 | 0.5302 | |
|
| 0.4081 | 551.72 | 4000 | 0.5190 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.38.1 |
|
- Pytorch 2.1.2 |
|
- Datasets 2.1.0 |
|
- Tokenizers 0.15.1 |
|
|
