voxtalkz

voxtalkz 是一个开源 python 库，旨在提供将类似播放的脚本转换为半复制音频剧的服务。

入门

先决条件

Python >= 3.6
pip

voxtalkz 需要几个 python 库才能运行：

Google Text To Speach (gTTS)

Pydub (pydub)

这些将使用 pip 自动安装。

安装

pip install voxtalkz

注意：如果pip失败，pip3应该可以工作

测试

编写一个简短的脚本并将其另存为test.script.

# Assign a name to an actor
Susan:british_woman

# Make her say Hello, World!
Susan:Hello, World!

有两个简单的选项可以运行 voxtalks

1. 打开命令行并导航到您的文件。

   运行脚本。

   python3 -m voxtalkz test.script test --debug
   

2. 在 python shell 中运行它：

   import voxtalks
   voxtalkz.say(<path/to/test.script>, <output_file_name>, debug=True)
   

假设一切顺利，您应该会看到一个名为 test.mp3 的文件。使用您喜欢的 mp3 播放程序播放它。（其他文件类型将根据要求提供）

使用 wavnet 语音

而不是python3 -m voxtalkz test.script test --debug, 使用python3 -m voxtalkz test.script test --debug --cloud <my_api_key>（可能不起作用）

或在空闲或脚本中： python import voxtalks voxtalkz.say(<path/to/test.script>, <output_file_name>, debug=True, cloudKey=<my_api_key>)

将 <my_api_key> 替换为 Google Cloud TextToSpeach API 密钥，（谷歌Google Cloud TextToSpeach API key它以了解如何获得）

这将为您提供比默认默认声音更好的声音，但请记住，wavenet 有不同的声音。查看帮助部分以获取更多信息。

通过 ssml 支持 wavenet 语音的修饰符，带有一个简单的转换脚本，例如：

Susan:real_american_woman

Susan: Hello World! __I really like dirt.__ *Dirt tasts good.*

听起来像：世界你好！我真的很喜欢泥土。 泥土味道不错。

请记住，强调一个词会破坏句子的流畅性，因此以下

Susan:real_american_woman

Susan: I __really__ like dirt.

听起来像我。真的。像泥土。

使用音效

导航到您的主目录（在 C:/Users/Username 的 Windows 上）并创建一个名为.voxtalkz.

导航到该文件夹​​并创建一个名为soundEffects.

下载 .mp3 或 .wav 音效（其他文件类型将根据要求提供，或根据我的需要提供）并将其放在.voxtalkz/soundEffects/文件夹中，对于本教程，我们假设文件名为footsteps.mp3.

编写一个简短的脚本并将其另存为test2.script.

# Assign a name to an actor
Susan:british_woman

# Make her say Hello, World!
Susan:Hello, World!
Susan:Goodbye, World!

# Add the soud effect
*footsteps

打开命令行并导航到您的文件。运行脚本。

python3 -m voxtalkz test2.script test2 --debug

播放 test2.mp3

添加修饰符

编写一个简短的脚本并将其另存为test3.script.

# Assign a name to an actor
Susan:british_woman

# Make her say Hello, World! as a little girl
Susan:Hello, World!@PITCH=0.3

# Make her say Goodbye, World! as an old woman
Susan:Goodbye, World!@PITCH=-0.3

打开命令行并导航到您的文件。运行脚本。

python3 -m voxtalkz test3.script test3 --debug

播放生成的 test3.mp3。

在 Python 中使用 in 和列表

	import voxtalkz

	TalkzBox = voxtalkz.voxTalkz('',<outputfilename>)
	mylist = [['Susan','british_woman'],['Susan', 'Hello, world!']]
	TalkzBox.ListToSound(mylist)

用作 Python 库

复制voxtalkz.py到包含要使用它的 python 文件的文件夹。

将这些代码行添加到您的程序中：

import voxtalkz

TalkzBox = voxtalkz.voxTalkz(<scriptname>,<outputfilename>)
Parsed = TalkzBox.Parse(TalkzBox.file)
OutputSound = TalkzBox.ListToSound(Parsed)
# OutputSound is an instance of pydub.AudioSegment. It can be proccesed with pydub now.
OutputSound.export(filename, format='<.wav, .mp3, and others>')

贡献

任何和所有的帮助将不胜感激！如果可能，将实施任何功能请求。

作者

• Don Flymoor -初始工作- DonFlymoor

执照

这个项目是在 GNUv3 许可证下获得许可的 - 请参阅LICENSE.md文件了解详细信息

用法

Usage: python3 -m voxtalkz [input file, output file] 

Converts play-like script to a .mp3 file 
Script file must be written in this manner: 

#The first time a unknown name is called, instead of making the person talk, the name will be assigned to a person. 
Susan:american_woman
#Then the person will "talk"
Susan:Hello, world!
#Comments are allowed!
*soundeffect_without_filename_extension

Effects can be applied by adding an @ symbal the the effect name, like so:
person1:hello, world!@VOLUME=8
A second effect can be applied by using the pipe("|") like so:
person1:Hello, World!@FADE|VOLUME=8

List of all effects:
	@FADE | Fade to nothing
	@FADE_IN | Fade in from silent
	@OVERLAY | Overlays the sound onto what has already been recorded. Use @OVERLAY=VAR1 to START the overlay at the begining of where you assigned @VAR=1
	@REPEAT= | Repeat audio segment however many times you specify. e.g. (american_woman:Hello, world!@REPEAT=10) would produce Someone saying "Hello, world!" ten times
	@VAR=    | Assign a number to a temporary table. Only used with @OVERLAY
	@VOLUME= | Set volume change in decibels. A negitive number will reduce the volume
	@PITCH=  | Set pitch change. e.g. "american_woman:Hello, world!@PITCH=0.3" would make the person sound like a little girl, while "american_woman:Hello, world!@PITCH=-0.3" would sound like an old woman
Wavenet/Cloud voice specific effects:
	@SPEAKINGRATE= | number from 0.25 to 4.0, where 1 is normal, 2 is twice as fast, 0.5 if half speed...

List of all normal actors:
	indian_man | Clearly speaks
	american_woman | Clearly speaks
	scottish_woman
	russian_woman
	drawling_man
	danish_woman | Clearly speaks
	dutch_woman
	au_woman | Clearly speaks
	british_woman | Clearly speaks
	indian_woman | Clearly speaks
	spanish_woman
	french_woman
	hindu_woman
	alien_man
	italian_man
	phone_woman
	korean_man | Clearly speaks
	dramatized_woman
	silly_man
	robot_man
	dramitized_girl
	bored_teen
	happy_girl
	boss_lady
	young_grandma
	spoiled_girl
	american_woman

List of all wavenet actors:
    "real_australian_woman":["en-AU-Wavenet-A","en-AU"],
    "real_australian_man":["en-AU-Wavenet-B","en-AU"],
    "real_personal_australian_woman":["en-AU-Wavenet-C","en-AU"],
    "real_personal_australian_man":["en-AU-Wavenet-D","en-AU"],

    "real_indian_woman":["en-IN-Wavenet-A","en-IN"],
    "real_indian_man":["en-IN-Wavenet-B","en-IN"],
    "real_personal_indian_woman":["en-IN-Wavenet-C","en-IN"],
    "real_personal_indian_man":["en-IN-Wavenet-D","en-IN"],

    "real_british_woman":["en-GB-Wavenet-A","en-GB"],
    "real_british_man":["en-GB-Wavenet-B","en-GB"],
    "real_personal_british_woman":["en-GB-Wavenet-C","en-GB"],
    "real_urgent_british_woman":["en-GB-Wavenet-F","en-GB"],
    "real_personal_british_man":["en-GB-Wavenet-D","en-GB"],

    "real_young_american_man":["en-US-Wavenet-A","en-US"],
    "real_middleage_american_man":["en-US-Wavenet-B","en-US"],
    "real_american_man":["en-US-Wavenet-B","en-US"],
    "real_middleage_american_woman":["en-US-Wavenet-C","en-US"],
    "real_american_woman":["en-US-Wavenet-C","en-US"],
    "real_middleage_personal_american_man":["en-US-Wavenet-D","en-US"],
    "real_middleage_personal_american_woman":["en-US-Wavenet-E","en-US"],
    "real_young_personal_american_woman":["en-US-Wavenet-F","en-US"],
    "real_distracted_middleage_american_woman":["en-US-Wavenet-G","en-US"],
    "real_young_american_woman":["en-US-Wavenet-H","en-US"],
    "real_young_personal_american_man":["en-US-Wavenet-I","en-US"],
    "real_cocky_american_man":["en-US-Wavenet-J","en-US"]
    + all normal actors

Sound effects must be in the .mp3 or .wav format and placed in /home/user/.voxtalk/soundEffects
To use footsteps.mp3 as a sound effect: put '*footsteps' in your script