Q: How do I use Azure Cognitive Services like Translator?

Q: How do I use Azure Cognitive Services like Translator?

A: There are many ways! I’ll show you some examples to get you started.

Cognitive Services refers to a wide variety of PaaS solutions that leverage AI. There are many services including speech to text, text to speech, sentiment analysis, conversational language understanding, translation, computer vision, Face API, anomaly detection, content moderator, and others. That is a lot of stuff and I’m not going to dive into all of it today but want to walk through how I’ve used Azure Translator today. Next week we’ll dive into how folks can deploy their own translator containers, then later we’ll cover how to make translator containers and text-analytics containers work together, then finally cover offline translation containers to cover the last of your edge cases and allowing you to leverage Azure without even connecting to Azure.

Before diving into deploying or using Translator in Azure folks might ask is: There are Azure Cognitive Services available in Azure so why would I want to have my own containers running cognitive services in the first place?

There are a few reasons and probably some that I’m haven’t thought of yet:

• My clients have poor network connectivity compared to the amount of data they need worked on so it would be too difficult to send and receive all my translations over the internet. This can also be the case for other cognitive services like sending large numbers of images for face detection over the internet.
  
• I have very latency sensitive workloads that are not near the cloud. For certain use cases like near real-time feedback to clients at remote sites leveraging computer vision they may want to minimize latency as much as possible by bringing processing from the cloud to their site. For workloads like these connected containers provide the best solution.
  
• There is no internet access at all on my client’s network. There are some clients like some research labs or government networks that have disconnected networks that don’t connect to the internet at all so they can’t directly query the cloud. For these workloads disconnected containers can leverage the power of Azure without directly connecting to Azure.
  

Those are all big problems for using the cloud but Microsoft is a cloud company now so it is nice that Azure can be leveraged no matter which unique situation your clients come up with.

Even if you want to use fancy containers for your production environment it makes sense to leverage Azure Translator in the cloud for dev or pre-prod. Testing in Azure is very easy to deploy without the need for resources to host containers and you can use the same code against Azure, connected containers, and disconnected containers. Testing in Azure is also great because there is a free tier so pre-production testing doesn’t cost much if anything extra.

All the documentation you could want for Translation in Azure is available here: Azure Cognitive Services Translator documentation - quickstarts, tutorials, API reference - Azure Cognitive Services

Finally, now we’re going to deploy and use Azure Translator the way nearly everyone will do it by leveraging Translator in Azure.

1. Log into the Azure portal then search for and click on the “Translators” service.
   

1. Select “Create” then fill in a few fields. I’m all about free stuff so it is nice that translation can be performed for free.
   

1. Click “Next” a few times then “Create” and now you have your Azure Translator resource.
   

That was almost too easy. Let’s try using our new resource a bit to make sure it works as expected.

We can see in the “Keys and Endpoint” section all the info we’ll need to do translations.

You can get all the information you need to write your own REST calls here: Translator V3.0 Reference - Azure Cognitive Services

I’ll provide a few examples here to show you how I like to do it with my favorite PowerShell and a little prettier output. I know a lot of you prefer curl, but there are already more examples covering curl in the official documentation, so I’ll show PowerShell here.

First, I initialize variables based on the “Keys and Endpoint” section my Translator resource. These variables are the only items you will need to change if you’re testing code in Azure, then running it later against connected or disconnected containers.

Disclaimer

The sample scripts are not supported under any Microsoft standard support program or service. The sample scripts are provided AS IS without warranty of any kind. Microsoft further disclaims all implied warranties including, without limitation, any implied warranties of merchantability or of fitness for a particular purpose. The entire risk arising out of the use or performance of the sample scripts and documentation remains with you. In no event shall Microsoft, its authors, or anyone else involved in the creation, production, or delivery of the scripts be liable for any damages whatsoever (including, without limitation, damages for loss of business profits, business interruption, loss of business information, or other pecuniary loss) arising out of the use of or inability to use the sample scripts or documentation, even if Microsoft has been advised of the possibility of such damages.

#Variable section

$baseURI = 'Azure Cognitive Services Translator documentation - quickstarts, tutorials, API reference - Azure Cognitive Services'

$Key = 'I_AM_NOT_SHARING_MY_KEY_BUT_PUT_YOURS_HERE'

$Region = 'eastus2'

$TextToTranslate = 'I like bananas and accordions.'

Next, I do a sample translation. If you know which language you’re translating from then you can specify “from” as documented in the API reference link above however if you do not specify it then it will automatically detect the source language for you.

#Test Translation:

$to = 'es'

$URI = $baseURI + 'translate?api-version=3.0&to=' + $to

$resp = Invoke-WebRequest -Uri $URI -Method Post -Headers $([hashtable]@{"Ocp-Apim-Subscription-Key"=$Key;"Ocp-Apim-Subscription-Region"=$Region}) -Body "[ {'Text':'$TextToTranslate'} ]" -ContentType "application/json; charset=utf-8" -ErrorAction Inquire

$trans = ($resp.Content | ConvertFrom-Json).translations.text

$trans

Me gustan los plátanos y los acordeones.

Next, I do language detection. Language detection is for I’m not sure what language I was provided and simply want to detect the language.

#Test language detection:

$URI = $baseURI + 'detect?api-version=3.0'

$resp = Invoke-WebRequest -Uri $URI -Method Post -Headers $([hashtable]@{"Ocp-Apim-Subscription-Key"=$Key;"Ocp-Apim-Subscription-Region"="EastUS2"}) -Body "[ {'Text':'$TextToTranslate'} ]" -ContentType "application/json; charset=utf-8" -ErrorAction Inquire

$trans = ($resp.Content | ConvertFrom-Json).language

$trans

en

Next, I list all the languages available. Listing the languages available is also helpful if you don’t recognize the information returned in language detection. If I didn’t know “en” meant English then I would just look it up in the table I output below. There are several languages where the “Native Name” is not output in a human readable format. This is not a concern for me since I don’t need those languages on this computer however if you have the right languages loaded they will display properly.

$URI = $baseURI + 'languages?api-version=3.0'

$resp = Invoke-WebRequest -Uri $URI -Method Get -Headers $([hashtable]@{"accept-language"="en"}) -ContentType "application/json; charset=utf-8" -ErrorAction Inquire

$trans = ($resp.Content | ConvertFrom-Json)

$Names = $trans.translation | Get-Member -MemberType NoteProperty

$Names.name | Select @{N='Abbreviation';E={$_}},@{N='Name';E={$trans.translation.($_).Name}},@{N='NativeName';E={$trans.translation.($_).NativeName}}

------------ ---- ----------

af Afrikaans Afrikaans

am Amharic አማáˆáŠ›

ar Arabic العربية

as Assamese à¦

সমীয়া

az Azerbaijani Azərbaycan

ba Bashkir Bashkir

bg Bulgarian БългарÑки

bn Bangla বাংলা

bo Tibetan བོད་སà¾à½‘་

bs Bosnian Bosnian

ca Catalan CatalÃ

cs Czech Čeština

cy Welsh Cymraeg

da Danish Dansk

de German Deutsch

dv Divehi Þ‹Þ¨ÞˆÞ¬Þ€Þ¨Þ„Þ¦ÞÞ°

el Greek Ελληνικά

en English English

es Spanish Español

et Estonian Eesti

eu Basque Euskara

fa Persian Ùارسی

fi Finnish Suomi

fil Filipino Filipino

fj Fijian Na Vosa Vakaviti

fo Faroese Føroyskt

fr French Français

fr-CA French (Canada) Français (Canada)

ga Irish Gaeilge

gl Galician Galego

gu Gujarati ગà«àªœàª°àª¾àª¤à«€

he Hebrew עברית

hi Hindi हिनà¥à¤¦à¥€

hr Croatian Hrvatski

hsb Upper Sorbian Hornjoserbšćina

ht Haitian Creole Haitian Creole

hu Hungarian Magyar

hy Armenian Õ€Õ¡ÕµÕ¥Ö€Õ¥Õ¶

id Indonesian Indonesia

ikt Inuinnaqtun Inuinnaqtun

is Icelandic Ãslenska

it Italian Italiano

iu Inuktitut áƒá“„ᒃᑎá‘ᑦ

iu-Latn Inuktitut (Latin) Inuktitut (Latin)

ja Japanese 日本語

ka Georgian ქáƒáƒ თული

kk Kazakh Қазақ Тілі

km Khmer ážáŸ’មែរ

kmr Kurdish (Northern) Kurdî (Bakur)

kn Kannada ಕನà³à²¨à²¡

ko Korean í•œêµì–´

ku Kurdish (Central) Kurdî (Navîn)

ky Kyrgyz Кыргызча

lo Lao ລາວ

lt Lithuanian Lietuvių

lv Latvian Latviešu

lzh Chinese (Literary) ä¸æ–‡ (文言文)

mg Malagasy Malagasy

mi MÄori Te Reo MÄori

mk Macedonian МакедонÑки

ml Malayalam മലയാളം

mn-Cyrl Mongolian (Cyrillic) Mongolian (Cyrillic)

mn-Mong Mongolian (Traditional) á ®á £á ©á á £á ¯ á ¬á ¡á ¯á ¡

mr Marathi मराठी

ms Malay Melayu

mt Maltese Malti

mww Hmong Daw Hmong Daw

my Myanmar (Burmese) မြန်မာ

nb Norwegian Norsk Bokmål

ne Nepali नेपाली

nl Dutch Nederlands

or Odia ଓଡ଼ିଆ

otq Querétaro Otomi Hñähñu

pa Punjabi ਪੰਜਾਬੀ

pl Polish Polski

prs Dari دری

ps Pashto پښتو

pt Portuguese (Brazil) Português (Brasil)

pt-PT Portuguese (Portugal) Português (Portugal)

ro Romanian Română

ru Russian РуÑÑкий

sk Slovak SlovenÄina

sl Slovenian SlovenÅ¡Äina

sm Samoan Gagana SÄmoa

so Somali Soomaali

sq Albanian Shqip

sr-Cyrl Serbian (Cyrillic) СрпÑки (ћирилица)

sr-Latn Serbian (Latin) Srpski (latinica)

sv Swedish Svenska

sw Swahili Kiswahili

ta Tamil தமிழà¯

te Telugu తెలà±à°—à±

th Thai ไทย

ti Tigrinya ትáŒáˆ

tk Turkmen Türkmen Dili

tlh-Latn Klingon (Latin) Klingon (Latin)

tlh-Piqd Klingon (pIqaD) Klingon (pIqaD)

to Tongan Lea Fakatonga

tr Turkish Türkçe

tt Tatar Татар

ty Tahitian Reo Tahiti

ug Uyghur ئۇيغۇرچە

uk Ukrainian УкраїнÑька

ur Urdu اردو

uz Uzbek (Latin) Uzbek (Latin)

vi Vietnamese Tiếng Việt

yua Yucatec Maya Yucatec Maya

yue Cantonese (Traditional) 粵語 (ç¹é«”)

zh-Hans Chinese Simplified ä¸æ–‡ (简体)

zh-Hant Chinese Traditional ç¹é«”ä¸æ–‡ (ç¹é«”)

zu Zulu Isi-Zulu

As you can see above some of the output doesn’t look right. I’m not fluent in these languages but some of that output doesn’t look correct. Uyghur, Cantonese, Chinese and others have a bunch of rectangles. This happened because the default output encoding for PowerShell 5 doesn’t properly display Unicode characters. Fortunately, if you run this in Windows Terminal or PowerShell 7 which I normally run in VS Code then characters are displayed properly as shown below.

Abbreviation Name NativeName

------------ ---- ----------

af Afrikaans Afrikaans

am Amharic አማርኛ

ar Arabic العربية

as Assamese অসমীয়া

az Azerbaijani Azərbaycan

ba Bashkir Bashkir

bg Bulgarian Български

bn Bangla বাংলা

bo Tibetan བོད་སྐད་

bs Bosnian Bosnian

ca Catalan Català

cs Czech Čeština

cy Welsh Cymraeg

da Danish Dansk

de German Deutsch

dv Divehi ދިވެހިބަސް

el Greek Ελληνικά

en English English

es Spanish Español

et Estonian Eesti

eu Basque Euskara

fa Persian فارسی

fi Finnish Suomi

fil Filipino Filipino

fj Fijian Na Vosa Vakaviti

fo Faroese Føroyskt

fr French Français

fr-CA French (Canada) Français (Canada)

ga Irish Gaeilge

gl Galician Galego

gu Gujarati ગુજરાતી

he Hebrew עברית

hi Hindi हिन्दी

hr Croatian Hrvatski

hsb Upper Sorbian Hornjoserbšćina

ht Haitian Creole Haitian Creole

hu Hungarian Magyar

hy Armenian Հայերեն

id Indonesian Indonesia

ikt Inuinnaqtun Inuinnaqtun

is Icelandic Íslenska

it Italian Italiano

iu Inuktitut ᐃᓄᒃᑎᑐᑦ

iu-Latn Inuktitut (Latin) Inuktitut (Latin)

ja Japanese 日本語

ka Georgian ქართული

kk Kazakh Қазақ Тілі

km Khmer ខ្មែរ

kmr Kurdish (Northern) Kurdî (Bakur)

kn Kannada ಕನ್ನಡ

ko Korean 한국어

ku Kurdish (Central) Kurdî (Navîn)

ky Kyrgyz Кыргызча

lo Lao ລາວ

lt Lithuanian Lietuvių

lv Latvian Latviešu

lzh Chinese (Literary) 中文 (文言文)

mg Malagasy Malagasy

mi Māori Te Reo Māori

mk Macedonian Македонски

ml Malayalam മലയാളം

mn-Cyrl Mongolian (Cyrillic) Mongolian (Cyrillic)

mn-Mong Mongolian (Traditional) ᠮᠣᠩᠭᠣᠯ ᠬᠡᠯᠡ

mr Marathi मराठी

ms Malay Melayu

mt Maltese Malti

mww Hmong Daw Hmong Daw

my Myanmar (Burmese) မြန်မာ

nb Norwegian Norsk Bokmål

ne Nepali नेपाली

nl Dutch Nederlands

or Odia ଓଡ଼ିଆ

otq Querétaro Otomi Hñähñu

pa Punjabi ਪੰਜਾਬੀ

pl Polish Polski

prs Dari دری

ps Pashto پښتو

pt Portuguese (Brazil) Português (Brasil)

pt-PT Portuguese (Portugal) Português (Portugal)

ro Romanian Română

ru Russian Русский

sk Slovak Slovenčina

sl Slovenian Slovenščina

sm Samoan Gagana Sāmoa

so Somali Soomaali

sq Albanian Shqip

sr-Cyrl Serbian (Cyrillic) Српски (ћирилица)

sr-Latn Serbian (Latin) Srpski (latinica)

sv Swedish Svenska

sw Swahili Kiswahili

ta Tamil தமிழ்

te Telugu తెలుగు

th Thai ไทย

ti Tigrinya ትግር

tk Turkmen Türkmen Dili

tlh-Latn Klingon (Latin) Klingon (Latin)

tlh-Piqd Klingon (pIqaD) Klingon (pIqaD)

to Tongan Lea Fakatonga

tr Turkish Türkçe

tt Tatar Татар

ty Tahitian Reo Tahiti

ug Uyghur ئۇيغۇرچە

uk Ukrainian Українська

ur Urdu اردو

uz Uzbek (Latin) Uzbek (Latin)

vi Vietnamese Tiếng Việt

yua Yucatec Maya Yucatec Maya

yue Cantonese (Traditional) 粵語 (繁體)

zh-Hans Chinese Simplified 中文 (简体)

zh-Hant Chinese Traditional 繁體中文 (繁體)

zu Zulu Isi-Zulu

These Azure cognitive services are very powerful and easy to use so they can be stitched together with a few REST calls to make some powerful applications. I look forward to seeing how more clients integrate these into their applications.

Have fun scripting!

References:

Cognitive Services documentation: Azure Cognitive Services documentation

Translator documentation: Azure Cognitive Services Translator documentation - quickstarts, tutorials, API reference - Azure Cognitive Services

Translator v3 API: Translator V3.0 Reference - Azure Cognitive Services

Continue reading...