File size: 1,285 Bytes
4e59d4c
cd3ba6d
 
 
 
 
 
 
 
 
 
 
4e59d4c
 
 
 
cd3ba6d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4e59d4c
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
---
tags:
- TensorFlowTTS
- audio
- text-to-speech
- text-to-mel
language: eng
license: apache-2.0
datasets:
- LJSpeech
widget:
- text: "How are you?"
---



This repository provides a pretrained [FastSpeech](https://arxiv.org/abs/1905.09263) trained on LJSpeech dataset (ENG). For a detail of the model, we encourage you to read more about
[TensorFlowTTS](https://github.com/TensorSpeech/TensorFlowTTS). 


## Install TensorFlowTTS
First of all, please install TensorFlowTTS with the following command:
```
pip install TensorFlowTTS
```

### Converting your Text to Mel Spectrogram
```python
import numpy as np
import soundfile as sf
import yaml

import tensorflow as tf

from tensorflow_tts.inference import AutoProcessor
from tensorflow_tts.inference import TFAutoModel

processor = AutoProcessor.from_pretrained("ruslanmv/tts-fastspeech-ljspeech-en")
fastspeech = TFAutoModel.from_pretrained("ruslanmv/tts-fastspeech-ljspeech-en")

text = "How are you?"

input_ids = processor.text_to_sequence(text)

mel_before, mel_after, duration_outputs = fastspeech.inference(
    input_ids=tf.expand_dims(tf.convert_to_tensor(input_ids, dtype=tf.int32), 0),
    speaker_ids=tf.convert_to_tensor([0], dtype=tf.int32),
    speed_ratios=tf.convert_to_tensor([1.0], dtype=tf.float32),
)
```