Skip to content

Commit 0196511

Browse files
JillGrant615JillGrant615
authored
Merge pull request #2415 from laujan/363370-translator-pro-ga
363370 translator pro ga
2 parents fd1c387 + f12ca17 commit 0196511

File tree

6 files changed

+381
-14
lines changed

6 files changed

+381
-14
lines changed

‎articles/ai-services/language-service/personally-identifiable-information/overview.md

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -18,11 +18,11 @@ PII detection is one of the features offered by [Azure AI Language](../overview.
1818

1919
## What's new
2020

21-
The Text PII and Conversational PII detection preview API (version `2024-11-15-preview`) now supports the option to mask detected sensitive entities with a label beyond just redaction characters. Customers have the option to specify if personally identifiable information content such as names and phone numbers, i.e. `“John Doe received a call from 424-878-9192”`, are masked with a redaction character, i.e. `“******** received a call from ************”`, or masked with an entity label, i.e. `“[PERSON_1] received a call from [PHONENUMBER_1]”`. More on how to specify the redaction policy style for your outputs can be found in our [how-to guides](how-to-call.md).
21+
The Text PII and Conversational PII detection preview API (version `2024-11-15-preview`) now supports the option to mask detected sensitive entities with a label beyond just redaction characters. Customers have the option to specify if personally identifiable information content such as names and phone numbers, i.e. `"John Doe received a call from 424-878-9192"`, are masked with a redaction character, i.e. `"******** received a call from ************"`, or masked with an entity label, i.e. `"[PERSON_1] received a call from [PHONENUMBER_1]"`. More on how to specify the redaction policy style for your outputs can be found in our [how-to guides](how-to-call.md).
2222

2323
The Conversational PII detection models (both version `2024-11-01-preview` and `GA`) have been updated to provide enhanced AI quality and accuracy. The numeric identifier entity type now also includes Drivers License and Medicare Beneficiary Identifier.
2424

25-
As of June 2024, we now provide General Availability support for the Conversational PII service (English-language only). Customers can now redact transcripts, chats, and other text written in a conversational style (i.e. text with “um”s, “ah”s, multiple speakers, and the spelling out of words for more clarity) with better confidence in AI quality, Azure SLA support and production environment support, and enterprise-grade security in mind.
25+
As of June 2024, we now provide General Availability support for the Conversational PII service (English-language only). Customers can now redact transcripts, chats, and other text written in a conversational style (i.e. text with "um"s, "ah"s, multiple speakers, and the spelling out of words for more clarity) with better confidence in AI quality, Azure SLA support and production environment support, and enterprise-grade security in mind.
2626

2727
> [!TIP]
2828
> Try out PII detection [in Azure AI Foundry portal](https://ai.azure.com/explore/language), where you can [utilize a currently existing Language Studio resource or create a new Azure AI Foundry resource](../../../ai-studio/ai-services/connect-ai-services.md)

‎articles/ai-services/translator/toc.yml

Lines changed: 5 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -284,6 +284,11 @@ items:
284284
- name: What is Microsoft Translator Pro?
285285
displayName: app service, mobile app, ios, speech-to-speech
286286
href: translator-pro/overview.md
287+
- name: Language support
288+
displayName: iso,language code, locale
289+
href: translator-pro/language-support.md
290+
- name: FAQ
291+
href: translator-pro/faq.yml
287292
- name: Responsible AI and compliance
288293
items:
289294
- name: Transparency note

‎articles/ai-services/translator/translator-pro/faq.yml

Lines changed: 120 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,120 @@
1+
### YamlMime:FAQ
2+
metadata:
3+
title: Microsoft Translator Pro FAQ
4+
description: Get answers to frequently asked questions about the Microsoft Translator Pro mobile application.
5+
author: laujan
6+
manager: nitinme
7+
ms.service: azure-ai-translator
8+
ms.custom: references_regions
9+
ms.topic: faq
10+
ms.date: 01/21/2025
11+
ms.author: lajanuar
12+
title: Frequently asked questions about Microsoft Translator Pro
13+
summary: |
14+
Microsoft Translator Pro is an advanced mobile application that enables seamless, real-time, speech-to-speech translation within the enterprise ecosystem.
15+
sections:
16+
- name: General
17+
questions:
18+
- question: |
19+
What is Microsoft Translator Pro?
20+
answer: |
21+
Microsoft Translator Pro is an advanced mobile application that enables seamless, real-time, speech-to-speech translation within the enterprise ecosystem. The app is designed to help organizations communicate effectively across languages and cultures, enabling employees to collaborate, share ideas, and engage with customers and partners in their preferred language.
22+
23+
- question: |
24+
How do I download the app?
25+
answer: |
26+
To use Microsoft Translator Pro, you must complete and submit the [Microsoft Translator Pro Limited Access](https://customervoice.microsoft.com/Pages/ResponsePage.aspx?id=v4j5cvGGr0GRqy180BHbR7en2Ais5pxKtso_Pz4b1_xUME5BUDBVWUlaNDlUN0FRODRIQ082SjFVUCQlQCN0PWcu) online request form for approval. Once you're approved, you receive an email with a link to download the app and onboarding instructions to onboard.
27+
28+
- question: |
29+
What platforms support Microsoft Translator Pro?
30+
answer: |
31+
Microsoft Translator Pro is available on the iOS platform and requires iOS 15 or newer versions.
32+
33+
- name: Data Management, Privacy, and Retention
34+
questions:
35+
- question: |
36+
What standard compliance frameworks and industry best practice measures are in place to ensure the security, confidentiality, and protection of personal data when using Microsoft Translator Pro?
37+
38+
answer: |
39+
Customer data submitted for translation to Microsoft Translator Pro is not stored permanently. There is no record of the submitted text, voice, or speech in any Microsoft data center. All organizational data remains within your organization’s Azure subscription, ensuring that no data is shared with Microsoft. You maintain full control over your data, which remains your exclusive business. This control is further strengthened by Microsoft’s adherence to extensive privacy laws, such as GDPR, and privacy standards, including the ISO/IEC 27018—the world's first international code of practice for cloud privacy. For more information,*see* [Microsoft Translator Confidentiality](https://www.microsoft.com/translator/business/notrace/#compliance).
40+
41+
- question: |
42+
Does Microsoft Translator Pro retain any user data for model improvement or share that data with developers for further training?
43+
answer: |
44+
No. For real-time speech-to-speech translation, no data is stored at rest. Microsoft doesn't retain data for model improvement.
45+
46+
- question: |
47+
Can our organization access the data generated by the app for further analysis or auditing purposes?
48+
answer: |
49+
Yes. You can use your generated data for your own analysis. Your data resides in your organization’s subscription. For more information, *see* [Microsoft Translator Confidentiality](https://aka.ms/notrace).
50+
51+
- question: |
52+
What privacy compliance frameworks does Microsoft Translator Pro follow?
53+
answer: |
54+
The Microsoft Translator Pro app adheres to the privacy policy and framework stated in the [Microsoft Privacy Statement](https://www.microsoft.com/privacy/privacystatement).
55+
56+
- name: Bias Mitigation and impact on equity and fairness
57+
questions:
58+
- question: |
59+
Does Microsoft implement measures to identify and mitigate potential bias in translation outputs, especially concerning minority languages, dialects, or culturally specific terminology?
60+
answer: |
61+
Yes. Microsoft actively researches and addresses bias-related issues. For more information, *see* [Azure AI Translator Transparency Note](/legal/cognitive-services/translator/transparency-note#bias).
62+
63+
- question: |
64+
Has the Microsoft Translator Pro app been evaluated for its impact on equity and fairness?
65+
answer: |
66+
Yes. Microsoft prioritizes creating technologies and products that are fair and inclusive. For more information, *see* [Azure AI Speech Transparency Note](/legal/cognitive-services/speech-service/speech-to-text/transparency-note#fairness-considerations).
67+
68+
- question: |
69+
Which demographic or cultural factors affect the accuracy or quality of translations?
70+
answer: |
71+
Microsoft Translator Pro is carefully trained to minimize biases and [supports multiple languages and accents](language-support.md). However, there can be variations in the quality and availability of voices across different languages. For more information, *see* [Transcription and translation accuracy](/legal/cognitive-services/translator/translator-pro-transparency-note#transcription-and-translation-accuracy).
72+
73+
- name: Training Data and Model Parameters
74+
questions:
75+
- question: |
76+
What data sources, metrics, and measures are used to train the models and ensure inclusive representation across different languages and cultural contexts?
77+
answer: |
78+
At Microsoft, our approach to managing the quality of translations is structured and ongoing. We ensure continuous improvement by implementing regular technology updates and biweekly release cycles. During model fine-tuning, we use Bilingual Evaluation Understudy (BLEU) and ChrF scores to identify potential release candidates. Before the final release, we also utilize COMET scores, which closely mirror human evaluations.
79+
80+
To ensure the highest quality, a release candidate must statistically outperform the current production model. Additionally, human judges assess new language models against reference translations. This process allows customers to enjoy the latest models without needing to change any code. For reported quality issues, we offer fast resolutions through a hotfix process for critical problems and address pattern-based issues through model updates.
81+
82+
- question: |
83+
Are models trained to handle information related to sensitive factors such as sex, gender, age, and other demographic characteristics to prevent discriminatory outputs or biases in translation quality?
84+
answer: |
85+
Yes, fairness is a complex social-technical factor that influences aspects of our product development. One dimension we consider is how well the system performs across groups based on factors such as race, ethnicity, region, gender, and age. For more information, *see* [Azure AI Translator transparency note](/legal/cognitive-services/translator/transparency-note#bias).
86+
87+
- name: Testing Standards, Evaluation, and Metrics
88+
questions:
89+
- question: |
90+
Are evaluation metrics and testing protocols implemented to ensure the accuracy and cultural sensitivity of translations, especially in critical areas like legal documents or emergency communications?
91+
answer: |
92+
Translation quality is always relative to a test set. We recommend that you measure quality for a representative test set for a given scenario. For more information, *see* [Azure AI Translator evaluation methods](/legal/cognitive-services/translator/transparency-note?context=%2Fazure%2Fai-services%2Ftranslator%2Fcontext%2Fcontext#evaluation-of-azure-ai-translator).
93+
94+
- question: |
95+
How does Microsoft Translator Pro ensure the quality and accuracy of translations in real-world, high-stakes environments?
96+
answer: |
97+
There are no universally recognized standard translation test sets for benchmarking and we refrain from publishing the absolute scores from our assessments of machine translation systems. Nevertheless, we continuously monitor quality and find that human evaluation yields the most significant and reliable results.
98+
99+
- question: |
100+
What measures are in place to ensure that translations are accurate and reliable in real-world scenarios?
101+
answer: |
102+
In addition to ongoing testing and evaluation, we also utilize speech and translation services trusted by customers for years in practical, real-world applications. For more information, *see* [Customer story: Berlitz empowered 500,000 language learners using Azure AI Speech](https://www.microsoft.com/customers/story/1650519804730300378-berlitz-language-azure-ai-speech-usa).
103+
104+
- name: Ongoing Monitoring, Accountability, and Transparency
105+
questions:
106+
- question: |
107+
Do you provide options for human review in cases of translation ambiguity or potential error?
108+
answer: |
109+
Microsoft Translator Pro doesn’t persist customer data submitted for translation and can't review your data. Your data resides within your subscription. You can access you data and perform human review on transcripts with the corresponding audio. For more information, *see* [Microsoft Translator Confidentiality](https://aka.ms/notrace).
110+
111+
- question: |
112+
What transparency measures are implemented to document and communicate model limitations, recurring issues, or risks associated with translations?
113+
answer: |
114+
Microsoft's Transparency Notes aim to clarify the workings of our AI technology, highlighting the choices available to system owners that can impact system performance and behavior. They emphasize the importance of considering the entire system, encompassing the technology, the people, and the environment. These notes can be a valuable resource when developing or deploying your own system. For more information, *see* [Microsoft Translator Pro Transparency Note](/legal/cognitive-services/translator/translator-pro-transparency-note#limitations).
115+
116+
- question: |
117+
How does the quality differ when the app utilizes on-device models versus online models?
118+
answer: |
119+
The on-device models utilized by the Microsoft Translator Pro app for offline processing differ from the cloud models, leading to a variation in quality. The models implemented on the device are smaller compared to models used in the cloud.
120+

‎articles/ai-services/translator/translator-pro/language-support.md

Lines changed: 136 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,136 @@
1+
---
2+
title: Language support - Microsoft Translator Pro
3+
titleSuffix: Azure AI services
4+
description: Language support for Microsoft Translator Pro speech-to-speech application.
5+
#services: cognitive-services
6+
author: laujan
7+
manager: nitinme
8+
ms.service: azure-ai-translator
9+
ms.topic: reference
10+
ms.date: 01/21/2025
11+
ms.author: lajanuar
12+
---
13+
14+
# Microsoft Translator Pro language support
15+
16+
Microsoft Translator Pro provides real-time multilingual speech-to-speech support. Our language support capabilities enable users to converse naturally and empower global communication. The following table lists the available language and locale support.
17+
18+
| Language | Online | On-device (doesn't require internet connection) |
19+
| --- | --- | --- |
20+
| Afrikaans (South Africa) | ✔️ | |
21+
| Albanian (Albania) | ✔️ | |
22+
| Amharic (Ethiopia) | ✔️ | |
23+
| Arabic (Algeria) | ✔️ | |
24+
| Arabic (Bahrain) | ✔️ | |
25+
| Arabic (Egypt) | ✔️ | |
26+
| Arabic (Iraq) | ✔️ | |
27+
| Arabic (Jordan) | ✔️ | |
28+
| Arabic (Kuwait) | ✔️ | |
29+
| Arabic (Lebanon) | ✔️ | |
30+
| Arabic (Libya) | ✔️ | |
31+
| Arabic (Morocco) | ✔️ | |
32+
| Arabic (Oman) | ✔️ | |
33+
| Arabic (Qatar) | ✔️ | |
34+
| Arabic (Saudi Arabia) | ✔️ | |
35+
| Arabic (Syria) | ✔️ | |
36+
| Arabic (Tunisia) | ✔️ | |
37+
| Arabic (United Arab Emirates) | ✔️ | |
38+
| Arabic (Yemen) | ✔️ | |
39+
| Armenian (Armenia) | ✔️ | |
40+
| Azerbaijani (Latin, Azerbaijan) | ✔️ | |
41+
| Basque | ✔️ | |
42+
| Bengali (India) | ✔️ | |
43+
| Bosnian (Bosnia and Herzegovina) | ✔️ | |
44+
| Bulgarian (Bulgaria) | ✔️ | |
45+
| Catalan (Spain) | ✔️ | |
46+
| Chinese (Cantonese, Traditional) | ✔️ | |
47+
| Chinese (Mandarin, Simplified) | ✔️ | ✔️ |
48+
| Chinese (Taiwanese Mandarin, Traditional) | ✔️ | |
49+
| Croatian (Croatia) | ✔️ | |
50+
| Czech (Czechia) | ✔️ | |
51+
| Danish (Denmark) | ✔️ | |
52+
| Dutch (Belgium) | ✔️ | |
53+
| Dutch (Netherlands) | ✔️ | |
54+
| Estonian (Estonia) | ✔️ | |
55+
| Filipino (Philippines) | ✔️ | |
56+
| Finnish (Finland) | ✔️ | |
57+
| French (Belgium) | ✔️ | |
58+
| French (Canada) | ✔️ | |
59+
| French (France) | ✔️ | ✔️ |
60+
| French (Switzerland) | ✔️ | |
61+
| Galician | ✔️ | |
62+
| Georgian (Georgia) | ✔️ | |
63+
| German (Austria) | ✔️ | |
64+
| German (Germany) | ✔️ | ✔️ |
65+
| German (Switzerland) | ✔️ | |
66+
| Greek (Greece) | ✔️ | |
67+
| Gujarati (India) | ✔️ | |
68+
| Hebrew (Israel) | ✔️ | |
69+
| Hindi (India) | ✔️ | |
70+
| Hungarian (Hungary) | ✔️ | |
71+
| Icelandic (Iceland) | ✔️ | |
72+
| Indonesian (Indonesia) | ✔️ | |
73+
| Irish (Ireland) | ✔️ | |
74+
| Italian (Italy) | ✔️ | ✔️ |
75+
| Japanese (Japan) | ✔️ | ✔️ |
76+
| Kannada (India) | ✔️ | |
77+
| Kazakh (Kazakhstan) | ✔️ | |
78+
| Khmer (Cambodia) | ✔️ | |
79+
| Kiswahili (Kenya) | ✔️ | |
80+
| Kiswahili (Tanzania) | ✔️ | |
81+
| Korean (Korea) | ✔️ | |
82+
| Lao (Laos) | ✔️ | |
83+
| Latvian (Latvia) | ✔️ | |
84+
| Lithuanian (Lithuania) | ✔️ | |
85+
| Macedonian (North Macedonia) | ✔️ | |
86+
| Malay (Malaysia) | ✔️ | |
87+
| Malayalam (India) | ✔️ | |
88+
| Maltese (Malta) | ✔️ | |
89+
| Marathi (India) | ✔️ | |
90+
| Mongolian (Mongolia) | ✔️ | |
91+
| Nepali (Nepal) | ✔️ | |
92+
| Norwegian Bokmal (Norway) | ✔️ | |
93+
| Pashto (Afghanistan) | ✔️ | |
94+
| Persian (Iran) | ✔️ | |
95+
| Polish (Poland) | ✔️ | |
96+
| Portuguese (Brazil) | ✔️ | |
97+
| Portuguese (Portugal) | ✔️ | |
98+
| Romanian (Romania) | ✔️ | |
99+
| Russian (Russia) | ✔️ | ✔️ |
100+
| Serbian (Cyrillic, Serbia) | ✔️ | |
101+
| Sinhala (Sri Lanka) | ✔️ | |
102+
| Slovak (Slovakia) | ✔️ | |
103+
| Slovenian (Slovenia) | ✔️ | |
104+
| Somali (Somalia) | ✔️ | |
105+
| Spanish (Argentina) | ✔️ | |
106+
| Spanish (Bolivia) | ✔️ | |
107+
| Spanish (Chile) | ✔️ | |
108+
| Spanish (Columbia) | ✔️ | |
109+
| Spanish (Costa Rica) | ✔️ | |
110+
| Spanish (Cuba) | ✔️ | |
111+
| Spanish (Dominican Republic) | ✔️ | |
112+
| Spanish (El Salvador) | ✔️ | |
113+
| Spanish (Equatorial Guinea) | ✔️ | |
114+
| Spanish (Guatemala) | ✔️ | |
115+
| Spanish (Honduras) | ✔️ | |
116+
| Spanish (Mexico) | ✔️ | ✔️ |
117+
| Spanish (Nicaragua) | ✔️ | |
118+
| Spanish (Panama) | ✔️ | |
119+
| Spanish (Paraguay) | ✔️ | |
120+
| Spanish (Peru) | ✔️ | |
121+
| Spanish (Puerto Rico) | ✔️ | |
122+
| Spanish (Spain) | ✔️ | |
123+
| Spanish (United States) | ✔️ | |
124+
| Spanish (Uruguay) | ✔️ | |
125+
| Spanish (Venezuela) | ✔️ | |
126+
| Swedish (Sweden) | ✔️ | |
127+
| Tamil (India) | ✔️ | |
128+
| Telugu (India) | ✔️ | |
129+
| Thai (Thailand) | ✔️ | |
130+
| Turkish (Türkiye) | ✔️ | |
131+
| Ukrainian (Ukraine) | ✔️ | |
132+
| Urdu (India) | ✔️ | |
133+
| Uzbek (Latin, Uzbekistan) | ✔️ | |
134+
| Vietnamese (Vietnam) | ✔️ | |
135+
| Welsh (United Kingdom) | ✔️ | |
136+
| isiZulu (South Africa) | ✔️ | |

0 commit comments

Comments
 (0)