Skip to content

Commit

Permalink
optimize tts feature
Browse files Browse the repository at this point in the history
  • Loading branch information
@cdhigh
cdhigh committed Apr 24, 2024
1 parent 9b4d37a commit f0f786a
Show file tree
Hide file tree
Showing 17 changed files with 744 additions and 487 deletions.
2 changes: 1 addition & 1 deletion application/lib/ebook_tts/engines/__init__.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@
from .azure import AzureTTS
from .google import GoogleWebTTSFree, GoogleTextToSpeech
builtin_tts_engines = {
AzureTTS.name: AzureTTS,
GoogleWebTTSFree.name: GoogleWebTTSFree,
GoogleTextToSpeech.name: GoogleTextToSpeech,
AzureTTS.name: AzureTTS,
}
1 change: 1 addition & 0 deletions application/lib/ebook_tts/engines/azure.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -207,6 +207,7 @@ class AzureTTS(TTSBase):
max_len_per_request = 1000
languages = azuretts_languages
regions = azure_regions
engine_url = 'https://azure.microsoft.com/en-us/products/ai-services/text-to-speech'
region_url = 'https://learn.microsoft.com/en-us/azure/ai-services/speech-service/regions'
voice_url = 'https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=tts'
language_url = 'https://learn.microsoft.com/en-us/azure/ai-services/speech-service/language-support?tabs=tts'
Expand Down
2 changes: 2 additions & 0 deletions application/lib/ebook_tts/engines/google.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -150,6 +150,7 @@ class GoogleWebTTSFree(TTSBase):
request_interval = 10 #额外的，好像每天只允许50个请求
max_len_per_request = 1666
languages = gtts_languages
engine_url = 'https://translate.google.com'

def __init__(self, params):
super().__init__(params)
Expand Down Expand Up @@ -184,6 +185,7 @@ class GoogleTextToSpeech(TTSBase):
request_interval = 2
max_len_per_request = 1666
languages = googletts_languages
engine_url = 'https://console.cloud.google.com/apis/api/texttospeech.googleapis.com/overview'

def __init__(self, params):
super().__init__(params)
Expand Down
1 change: 1 addition & 0 deletions application/lib/ebook_tts/engines/tts_base.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@ class TTSBase:
max_len_per_request = 500
languages = {}
regions = {}
engine_url = '' #一个链接，关于引擎的介绍链接网页
region_url = '' #一个链接，可以在这个链接网页上找到可用的区域
voice_url = '' #一个链接，可以在这个网页上找到语音名称列表
language_url = '' #一个链接，可以在这个网页上找到支持的语种列表
Expand Down
6 changes: 3 additions & 3 deletions application/lib/ebook_tts/html_audiolator.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -11,9 +11,9 @@ def get_tts_engines():
for name, engine in builtin_tts_engines.items():
info[name] = {'alias': engine.alias, 'need_api_key': engine.need_api_key,
'default_api_host': engine.default_api_host, 'api_key_hint': engine.api_key_hint,
'languages': engine.languages, 'region_url': engine.region_url,
'voice_url': engine.voice_url, 'language_url': engine.language_url,
'regions': engine.regions}
'languages': engine.languages, 'engine_url': engine.engine_url,
'region_url': engine.region_url, 'voice_url': engine.voice_url,
'language_url': engine.language_url, 'regions': engine.regions}
return info

class HtmlAudiolator:
Expand Down
175 changes: 104 additions & 71 deletions application/lib/pymp3cat.py
View file
Original file line number Diff line number Diff line change
@@ -1,8 +1,11 @@
#!/usr/bin/env python3
# -*- coding:utf-8 -*-
#Author: cdhigh<https://github.com/cdhigh>
#合并mp3文件
#来源：https://github.com/dmulholl/mp3cat
#将go语言转换为python，方便类似GAE这样不能执行二进制文件的平台合并mp3
#原始来源：https://github.com/dmulholl/mp3cat
#将go语言转换为python，没有第三方依赖，方便类似GAE这样不能执行二进制文件的平台合并mp3
#因为不需要解码，只是将源文件中的音乐帧逐个拷贝到目标文件，
#速度尽管慢一些（大概是go版本的2到3倍时间），在可以接受的范围内
import os, io, struct

#版本
Expand All @@ -21,21 +24,21 @@
DualChannel = 2
Mono = 3

#位率对应表
v1_br = {
3: (0, 32, 64, 96, 128, 160, 192, 224, 256, 288, 320, 352, 384, 416, 448), #layer1
2: (0, 32, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384), #layer2
1: (0, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320), #layer3
}
#位率对应表[layer][bitRateIndex]
v1_br = [(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0),
(0, 32000, 40000, 48000, 56000, 64000, 80000, 96000, 112000, 128000, 160000, 192000, 224000, 256000, 320000, 0), #layer3
(0, 32000, 48000, 56000, 64000, 80000, 96000, 112000, 128000, 160000, 192000, 224000, 256000, 320000, 384000, 0), #layer2
(0, 32000, 64000, 96000, 128000, 160000, 192000, 224000, 256000, 288000, 320000, 352000, 384000, 416000, 448000, 0), #layer1
]

v2_br = {
3: (0, 32, 48, 56, 64, 80, 96, 112, 128, 144, 160, 176, 192, 224, 256),
2: (0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160),
1: (0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160),
}
v2_br = [(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0),
(0, 8000, 16000, 24000, 32000, 40000, 48000, 56000, 64000, 80000, 96000, 112000, 128000, 144000, 160000, 0),
(0, 8000, 16000, 24000, 32000, 40000, 48000, 56000, 64000, 80000, 96000, 112000, 128000, 144000, 160000, 0),
(0, 32000, 48000, 56000, 64000, 80000, 96000, 112000, 128000, 144000, 160000, 176000, 192000, 224000, 256000, 0),
]

#采样率对应表[ver][layer]
samplingTable = ((11025, 12000, 8000), (0, 0, 0), (22050, 24000, 16000), (44100, 48000, 32000))
#采样率对应表[ver][samplingRateIndex]
samplingTable = ((11025, 12000, 8000, 0), (0, 0, 0, 0), (22050, 24000, 16000, 0), (44100, 48000, 32000, 0))

#每帧的采样数对应表[ver][layer]
sampleCountTable = ((0, 576, 1152, 384), (0, 0, 0, 0), (0, 576, 1152, 384), (0, 1152, 1152, 384))
Expand Down Expand Up @@ -77,7 +80,7 @@ def NextObject(stream):
#ID3v1标识: 'TAG'，包括标签头在内，一共128字节
if header1[0:3] == b'TAG':
stream.seek(start + 128)
return {'type': 'TAG', 'start': start, 'end': stream.tell(), 'len': 128}
return {'type': 'TAG', 'len': 128} #'start': start, 'end': stream.tell(),
elif header1[0:3] == b'ID3':
#ID3V2头一共10个字节
#char Header[3]; #ID3
Expand All @@ -94,17 +97,17 @@ def NextObject(stream):

stream.seek(start)
frame = {'type': 'ID3', 'len': length}
frame['start'] = start
#frame['start'] = start
frame['raw'] = stream.read(length + 10) #长度不包含头部10个字节
frame['end'] = stream.tell()
#frame['end'] = stream.tell()
return frame
elif (header1[0] == 0xff) and ((header1[1] & 0xe0) == 0xe0): #11比特的1，一个音乐数据帧开始
frame = ParseMusicHeader(header1)
if frame:
stream.seek(start)
frame['start'] = start
#frame['start'] = start
frame['raw'] = stream.read(frame['len']) #帧长度包含头部4个字节
frame['end'] = stream.tell()
#frame['end'] = stream.tell()
return frame

#出错，往后跳一个字节再重新尝试
Expand Down Expand Up @@ -153,38 +156,32 @@ def ParseMusicHeader(header):
if mpegVer == MPEGVersionReserved:
return None
layer = (header[1] & 0x06) >> 1 #2位，层, 0-未使用，1-Layer3, 2-Layer2, 3-Layer3
if layer == 0:
return None
crcProt = (header[1] & 0x01) == 0x00 #是否有CRC校验，0-校验
#crcProt = (header[1] & 0x01) == 0x00 #是否有CRC校验，0-校验
bitRateIndex = (header[2] & 0xf0) >> 4 #位率索引，共4位
if bitRateIndex == 0 or bitRateIndex == 15:
return None

#查表得出位率
if mpegVer == MPEGVersion1:
bitRate = v1_br.get(layer)[bitRateIndex] * 1000
bitRate = v1_br[layer][bitRateIndex] #查表得出位率
else:
bitRate = v2_br.get(layer)[bitRateIndex] * 1000
bitRate = v2_br[layer][bitRateIndex]
if bitRate == 0:
return None

samplingRateIndex = (header[2] & 0x0c) >> 2 #采样率索引，2位
if samplingRateIndex == 3:
samplingRate = samplingTable[mpegVer][samplingRateIndex] #查表得出采样率
if samplingRate == 0:
return None

#查表得出采样率
samplingRate = samplingTable[mpegVer][samplingRateIndex]

paddingBit = (header[2] & 0x02) == 0x02 #帧长调节 (1 bit)
privateBit = (header[2] & 0x01) == 0x01 #保留字 (1 bit)
#privateBit = (header[2] & 0x01) == 0x01 #保留字 (1 bit)
channelMode = (header[3] & 0xc0) >> 6 #声道模式 (2 bits)
modeExtension = (header[3] & 0x30) >> 4 #扩充模式，仅用于 Joint Stereo mode. (2 bits)
if (channelMode != JointStereo) and (modeExtension != 0):
return None

copyrightBit = (header[3] & 0x08) == 0x08 #版权 (1 bit)
originalBit = (header[3] & 0x04) == 0x04 #原版标志 (1 bit)
emphasis = (header[3] & 0x03) #强调标识 (2 bits)
if emphasis == 2:
return None
#modeExtension = (header[3] & 0x30) >> 4 #扩充模式，仅用于 Joint Stereo mode. (2 bits)
#if (channelMode != JointStereo) and (modeExtension != 0):
# return None

#copyrightBit = (header[3] & 0x08) == 0x08 #版权 (1 bit)
#originalBit = (header[3] & 0x04) == 0x04 #原版标志 (1 bit)
#emphasis = (header[3] & 0x03) #强调标识 (2 bits)
#if emphasis == 2:
# return None

#帧大小即每帧的采样数，表示一帧数据中采样的个数
sampleCount = sampleCountTable[mpegVer][layer]
Expand All @@ -208,8 +205,12 @@ def ParseMusicHeader(header):
# Experimentation on mp3 files captured from the wild indicates that it
# includes the header at least.
frameLength = int((sampleCount / 8) * bitRate / samplingRate + padding)
return {'type': 'FRAME', 'len': frameLength, 'bitRate': bitRate, 'samplingRate': samplingRate,
'sampleCount': sampleCount, 'mpegVer': mpegVer, 'layer': layer, 'channelMode': channelMode}

#每帧持续时间(毫秒) = 每帧采样数 / 采样频率 * 1000
#duration = sampleCount / samplingRate * 1000

return {'type': 'FRAME', 'len': frameLength, 'bitRate': bitRate,
'mpegVer': mpegVer, 'layer': layer, 'channelMode': channelMode}

#创建一个新的VBR帧
def NewXingHeader(totalFrames, totalBytes):
Expand Down Expand Up @@ -274,38 +275,46 @@ def AddID3v2Tag(output, input_):

#合并mp3文件
#output: 输出文件名或流对象
#inputs: 输入文件名列表或二进制内容类别
#inputs: 输入文件名列表或二进制内容列表
#tagIndex: 是否需要将第n个文件的ID3拷贝过来
#useBuffer: 是否使用内存缓冲区，仅仅适用于传入的参数是文件名
#force: 是否覆盖目标文件
#quiet: 是否打印过程
def merge(output: str, inputs: list, tagIndex: int=None, force: bool=True, quiet: bool=False):
#返回合并的文件数
def merge(output: str, inputs: list, tagIndex=None, useBuffer=True, force=False, quiet=False):
if not force and isinstance(output, str) and os.path.exists(output):
print(f"Error: the file '{output}' already exists.")
return
return 0
if inputs and isinstance(inputs[0], str) and output in inputs:
print(f'Error: the list of input files includes the output file.')
return
return 0

printInfo = (lambda x: x) if quiet else (lambda x: print(x))

outputStream = open(output, 'wb') if isinstance(output, str) else output
if isinstance(output, str):
outputStream = io.BytesIO() if useBuffer else open(output, 'wb')
else:
outputStream = output

totalFrames = 0
totalBytes = 0
totalFiles = 0
firstBitRate = 0
isVBR = False
for idx, input_ in enumerate(inputs):
needClose = False
if isinstance(input_, str):
printInfo(f' + {input_}')
input_ = open(input_, 'rb')
needClose = True
if useBuffer:
with open(input_, 'rb') as f:
inputStream = io.BytesIO(f.read())
else:
inputStream = open(input_, 'rb')
else:
printInfo(f' + <stream {idx}>')
inputStream = input_

isFirstFrame = True
for frame in IterFrame(input_):
for frame in IterFrame(inputStream):
if isFirstFrame: #第一个帧如果是VBR，不包含音乐数据
isFirstFrame = False
if IsVBRHeader(frame):
Expand All @@ -320,34 +329,58 @@ def merge(output: str, inputs: list, tagIndex: int=None, force: bool=True, quiet
totalFrames += 1
totalBytes += frame['len']
totalFiles += 1
if needClose:
input_.close()

if isinstance(input_, str) and not useBuffer:
inputStream.close()

if isinstance(output, str):
outputStream.close()
if useBuffer:
outputStream.seek(0)
else:
outputStream.close()

#如果不同的文件的比特率不同，则在前面添加一个VBR头
if isVBR:
printInfo("• Multiple bitrates detected. Adding VBR header.")
AddXingHeader(output, totalFrames, totalBytes)
if isinstance(output, str):
try:
tempStream.close()
os.remove(output + '.mp3cat.tmp')
except:
pass

if tagIndex is not None and tagIndex < len(inputs):
if isinstance(output, str) and not useBuffer:
AddXingHeader(output, totalFrames, totalBytes)
else:
AddXingHeader(outputStream, totalFrames, totalBytes)

#拷贝ID3Tag
if tagIndex is not None and (0 < tagIndex < len(inputs)):
input_ = inputs[tagIndex]
needClose = False
if isinstance(input_, str):
printInfo(f"• Copying ID3 tag from: {input_}")
input_ = open(input_, 'rb')
needClose = True
if useBuffer:
with open(input_, 'rb') as f:
inputStream = io.BytesIO(f.read())
else:
inputStream = open(input_, 'rb')
else:
printInfo(f'• Copying ID3 tag from: <stream {tagIndex}>')
AddID3v2Tag(output, input_)
if needClose:
input_.close()
inputStream = input_
if isinstance(output, str) and not useBuffer:
AddID3v2Tag(output, inputStream)
else:
AddID3v2Tag(outputStream, inputStream)
if isinstance(input_, str) and not useBuffer:
inputStream.close()

if isinstance(output, str) and useBuffer:
with open(output, 'wb') as f:
f.write(outputStream.getvalue())

printInfo(f"• {totalFiles} files merged.")
return totalFiles

if __name__ == '__main__':
import time
dir_ = 'd:/temp'
inputs = [os.path.join(dir_, f'Headspace{idx}.mp3') for idx in range(1, 9)]
inputs.append('d:/temp/na.mp3')
output = os.path.join(dir_, 'output.mp3')
startTime = time.time()
merge(output, inputs, force=True, useBuffer=True, tagIndex=8)
print(f'Consumed time: {time.time() - startTime:0.1f} seconds')
37 changes: 17 additions & 20 deletions application/static/base.js
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1362,27 +1362,24 @@ function TTSEngineFieldChanged(language, region) {
$('#tts_region_div').hide();
}

let setTtsHref = function (target_id, url) {
let target = $(target_id);
if (url) {
target.attr('href', url);
target.removeAttr('onclick');
target.css('text-decoration', 'underline dotted');
} else {
target.attr('href', 'javascript:void(0)');
target.attr('onclick', 'return false;');
target.css('text-decoration', 'none');
}
};

//设置提示的链接
let region_a = $('#tts_region_a');
if (engine.region_url) {
region_a.attr('href', engine.region_url);
region_a.removeAttr('onclick');
region_a.css('text-decoration', 'underline dotted');
} else {
region_a.attr('href', 'javascript:void(0)');
region_a.attr('onclick', 'return false;');
region_a.css('text-decoration', 'none');
}
let voice_a = $('#tts_voice_a');
if (engine.voice_url) {
voice_a.attr('href', engine.voice_url);
voice_a.removeAttr('onclick');
voice_a.css('text-decoration', 'underline dotted');
} else {
voice_a.attr('href', 'javascript:void(0)');
voice_a.attr('onclick', 'return false;');
voice_a.css('text-decoration', 'none');
}
setTtsHref('#tts_engine_a', engine.engine_url);
setTtsHref('#tts_region_a', engine.region_url);
setTtsHref('#tts_voice_a', engine.voice_url);
setTtsHref('#tts_language_a', engine.language_url);

//更新语种代码
let tts_language_sel = $('#tts_language_sel');
Expand Down
Loading

0 comments on commit f0f786a

Please sign in to comment.