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: https://learn.microsoft.com/en-us/azure/cognitive-services/translator/
Finally, now we’re going to deploy and use Azure Translator the way nearly everyone will do it by leveraging Translator in Azure.
- Log into the Azure portal then search for and click on the “Translators” service.
- 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.
- 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: https://learn.microsoft.com/en-us/azure/cognitive-services/translator/reference/v3-0-reference
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 = 'https://api.cognitive.microsofttranslator.com/'
$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: https://learn.microsoft.com/en-us/azure/cognitive-services/
Translator documentation: https://learn.microsoft.com/en-us/azure/cognitive-services/translator/
Translator v3 API: https://learn.microsoft.com/en-us/azure/cognitive-services/translator/reference/v3-0-reference
Microsoft
