Hi all,
I'm happy to announce that SDL Machine Translation Edge 8.5 in going to be released shortly.
SDL Machine Translation Edge 8.5 is a significant release which is introducing multiple new features and improvements, including Real-time adaptation, a new UI design, Kubernetes support and more.
What's new?
- Smart MT - Real-time adaptation
- Users can now submit feedback in real-time in the Translate User Interface
- Feedback can be reviewed by administrators and once approved, applied automatically to adapt the Machine Translation output
- User Interface Improvements
- Filter and Search capabilities have been added to Translation Queue, Dictionaries and Dictionaries segments
- Ability to select multiple translation jobs in the Translation Queue and download them all at once
- The MT Edge UI is now aligned with MT Cloud: the navigation menu is now displayed at the top
- Label support
- It is now possible to assign label(s) to translation jobs. Translations can be filtered and searched based on specific labels.
- In the Translate UI, it is possible to define default labels which will be applied to all document translations performed
- In the Translation Queue, it is possible to select multiple translation jobs and apply label(s) to the entire selection
- Enhanced PDF support with ABBYY OCR
- A new OCR option with ABBYY is now available to improve support of PDF
- Improved support for Scanned PDF
- Improved support for Arabic, Chinese, Japanese, Korean, Hebrew, Farsi, Thai and Vietnamese
- This option must be licensed specifically and 2 options are available:
- "ABBYY Enhanced" covering languages based on Roman, Cyrillic or Greek alphabet
- "ABBYY Enhanced+" covering "standard" languages + Arabic, Chinese, Japanese, Korean, Hebrew, Farsi, Thai and Vietnamese
- A new OCR option with ABBYY is now available to improve support of PDF
- Language Identification can now be configured to select the scope of language identification: All languages, Only SDL supported languages, Only installed languages or Only running languages
- Language Identification can be fully disabled in the UI. When disabled, auto-detect is not available in the Translate UI.
- Kubernetes support
- MT Edge can now be deployed on a Kubernetes cluster, using provided Helm charts.
- When deploying MT Edge on a Kubernetes cluster, the benefits include:
- Simplified Configuration of the MT Edge application using yaml files
- Automatic deployment of MT Edge Controller, job engine(s) and translation engine(s) as pods
- Job and Translation engines are available as statefulsets and can be scaled manually or automatically, based on needs.
- Auto-scaling allows you to define the minimum and maximum number of pods that can be started for a specific language pair, and what CPU usage threshold must be met to start a new pod. Pods are then automatically started based on current workload. This feature can be combined with Kubernetes nodes auto-scaling and ensure nodes are only started when more pods needs to be deployed
- Ability to send documents for Human Review by email directly from the MT Edge User Interface
- When SMTP and the address email of a reviewer is configured in MT Edge, documents can be sent with 1 click for Human review and Translation.
- Administrator can define the email address of the reviewer contact and define an email template which will be used when sending the file automatically
Enhancements
- Performance improvements
- Hardware requirements: when running Neural models on CPU, the CPU core requirements have been reduced by 50%
- When using CPU Optimized for Quality, the minimum requirements are now 2 CPU cores to reach 2000 wpm (1PU)
- When using CPU Optimized for Speed, the minimum requirements are now 1 CPU core to reach 2000 wpm (1PU)
- SAML configuration now supports Roles in addition to Groups. Okta is also officially supported as an Identity Provider.
- The Reporting UI now includes the ability to filter the statistics based on a date range, on users and on languages.
- Security enhancements
- The password policy now requires a minimum of 8 characters and it must include 3 of the following categories: Uppercase characters, Lowercase characters, Number and Special characters.
- Downloading a translation result from the Translation Queue is now only allowed for the user who has submitted the translation. Administrator will be able to view all translation jobs in the Translation Queue, but won't be able to download the resulting translation.
- Ability to adjust the MT Edge components configuration (job engines, translation engines, training engine) through API.
- It is no longer required to change configuration files manually and all parameters can be changed on all relevant engines at once using API.
- GPU support
- Nvidia TitanV is now supported (for Translation only)
- Nuance Translation Engine (NTE) is now supported up to version 4.7 for Audio transcription.
- The "master host" terminology is replaced by "controller host"
- The "Quick Translate" tab is renamed "Translate"
The online documentation has already been updated and can be accessed on our documentation server at https://docs.sdl.com
Our download servers will be updated as soon as the release is generally available, and we will notify this forum once done.
We hope you'll enjoy this new release, and please, let us know if you have any question,
Best regards,
Arnaud