iT邦幫忙

2021 iThome 鐵人賽

DAY 14
0

Azure cognitive service: Text-to-Speech- Azure 唸給你聽

Text-to-Speech 文字轉換語音是 Azure Speech 語音服務中其中一個項目。在語音服務之中,除了提供文字和語音的互相轉換以外,還有一些更進階的服務,如語音翻譯、從話語中理解發話者的意圖、辨識發話者身份,等等。在這次要執行的 Text-to-Speech 是為了把之前用 OCR 辨識到的韓文,直接轉換成語音,讓使用者可以不但可以從翻譯知道意思,也可以學習發音。

申請 Azure Speech

  1. 進入https://portal.azure.com/#home 。
  2. 點選建立資源
  3. 搜尋並選擇 speech 語音
  4. 自己命名。
  5. 找到可以選擇定價層 Free F0 的區域,並選擇 Free F0 。
  6. 給予標籤。
  7. 檢閱 + 建立。

安裝python套件

pip3.7 install azure-cognitiveservices-speech

金鑰與端點

  • 進入https://portal.azure.com/#home 。
  • 點選所有資源。
  • 點選剛剛建立的 Speech。
  • 點選金鑰與端點
  • 複製金鑰與端點

示範程式

一開始需要設定 speech config,輸入金鑰和當初申請時所選擇的區域代碼,會將字串傳給 Speech 轉換成語音,語音內容可以在本地端存成wav檔。
關於選擇區域的代碼,可以執行az account list-locations -o table,會列出所有區域相對應的代碼,在這邊我選擇的是美國東部 2 ,對應到的代碼就是eastus2。下面有列出對照表可供參考。
另外,這裡也需要設定發音的語系,關於語系設定的代碼,可以從此語言支援文件查詢。


from azure.cognitiveservices.speech import (
    SpeechConfig,
    SpeechSynthesizer,
)

# 輸入金鑰和區域,設定config
SPEECH_KEY = "你的金鑰"
speech_config = SpeechConfig(subscription=SPEECH_KEY, region="eastus2")

# 因為後續要針對韓文發音,所以設定成"ko-KR"
speech_config.speech_synthesis_language = "ko-KR"

# 設定轉換後存檔路徑
file_name = "your_path.wav"
audio_config = AudioOutputConfig(filename=file_name)

synthesizer = SpeechSynthesizer(
    speech_config=speech_config, audio_config=audio_config
)

# 開始轉換成語音
synthesizer.speak_text_async(string)

區域代碼表

這邊列出執行az account list-locations -o table之後,會出現的區域代碼表,方便大家查詢。

DisplayName Name RegionalDisplayName
East US eastus (US) East US
East US 2 eastus2 (US) East US 2
South Central US southcentralus (US) South Central US
West US 2 westus2 (US) West US 2
West US 3 westus3 (US) West US 3
Australia East australiaeast (Asia Pacific) Australia East
Southeast Asia southeastasia (Asia Pacific) Southeast Asia
North Europe northeurope (Europe) North Europe
Sweden Central swedencentral (Europe) Sweden Central
UK South uksouth (Europe) UK South
West Europe westeurope (Europe) West Europe
Central US centralus (US) Central US
North Central US northcentralus (US) North Central US
West US westus (US) West US
South Africa North southafricanorth (Africa) South Africa North
Central India centralindia (Asia Pacific) Central India
East Asia eastasia (Asia Pacific) East Asia
Japan East japaneast (Asia Pacific) Japan East
Jio India West jioindiawest (Asia Pacific) Jio India West
Korea Central koreacentral (Asia Pacific) Korea Central
Canada Central canadacentral (Canada) Canada Central
France Central francecentral (Europe) France Central
Germany West Central germanywestcentral (Europe) Germany West Central
Norway East norwayeast (Europe) Norway East
Switzerland North switzerlandnorth (Europe) Switzerland North
UAE North uaenorth (Middle East) UAE North
Brazil South brazilsouth (South America) Brazil South
Central US (Stage) centralusstage (US) Central US (Stage)
East US (Stage) eastusstage (US) East US (Stage)
East US 2 (Stage) eastus2stage (US) East US 2 (Stage)
North Central US (Stage) northcentralusstage (US) North Central US (Stage)
South Central US (Stage) southcentralusstage (US) South Central US (Stage)
West US (Stage) westusstage (US) West US (Stage)
West US 2 (Stage) westus2stage (US) West US 2 (Stage)
Asia asia Asia
Asia Pacific asiapacific Asia Pacific
Australia australia Australia
Brazil brazil Brazil
Canada canada Canada
Europe europe Europe
Global global Global
India india India
Japan japan Japan
United Kingdom uk United Kingdom
United States unitedstates United States
East Asia (Stage) eastasiastage (Asia Pacific) East Asia (Stage)
Southeast Asia (Stage) southeastasiastage (Asia Pacific) Southeast Asia (Stage)
Central US EUAP centraluseuap (US) Central US EUAP
East US 2 EUAP eastus2euap (US) East US 2 EUAP
West Central US westcentralus (US) West Central US
South Africa West southafricawest (Africa) South Africa West
Australia Central australiacentral (Asia Pacific) Australia Central
Australia Central 2 australiacentral2 (Asia Pacific) Australia Central 2
Australia Southeast australiasoutheast (Asia Pacific) Australia Southeast
Japan West japanwest (Asia Pacific) Japan West
Jio India Central jioindiacentral (Asia Pacific) Jio India Central
Korea South koreasouth (Asia Pacific) Korea South
South India southindia (Asia Pacific) South India
West India westindia (Asia Pacific) West India
Canada East canadaeast (Canada) Canada East
France South francesouth (Europe) France South
Germany North germanynorth (Europe) Germany North
Norway West norwaywest (Europe) Norway West
Sweden South swedensouth (Europe) Sweden South
Switzerland West switzerlandwest (Europe) Switzerland West
UK West ukwest (Europe) UK West
UAE Central uaecentral (Middle East) UAE Central
Brazil Southeast brazilsoutheast (South America) Brazil Southeast

現在已經具備 OCR、翻譯和文字轉換語音的工具了,所以下一篇我們來組裝韓文翻譯機器人。(謎之聲:搜集完武器,要打第二個中頭目了~)


上一篇
Day 13 Azure cognitive service: Translator- 讓 Azure 幫你翻譯
下一篇
Day 15 Chatbot integration- 韓文翻譯機器人
系列文
我不太懂 AI,可是我會一點 Python 和 Azure30
圖片
  直播研討會
圖片
{{ item.channelVendor }} {{ item.webinarstarted }} |
{{ formatDate(item.duration) }}
直播中

尚未有邦友留言

立即登入留言