What's new
This change log provides a summary of new features and enhancements for the Nuance Mix platform.
More Info:
See the Mix Learn tab for feature announcements and other improvements to the platform2024-06-18
Documentation update
The ASRaaS service has been updated with these enhancements:
-
A new result type, NOTIFICATIONS, returns information when the ASR instance is about to expire. This is useful if you have long audio sessions, which may terminate due to the normal operation of the underlying Kubernetes pods. See EnumResultType. No changes to your client applications are required.
-
The result.proto file includes the new notifications feature. See gRPC setup to download this file and generate client stubs if required. If you don’t need this feature, you may continue to use your existing proto file.
-
A new status code (512: Shutdown while processing a request) is emitted when the service closes. See Status codes.
2024-05-09
Documentation update
In ASRaaS, the documentation for the sample recognition client explains how to use languages other than American English. See Sample recognition client. The sample client itself was not updated.
In DLGaas, the sample client now accepts speech input as an audio file and can generate synthesized speech output as audio files. See Sample Python runtime client.
2024-03-27
Enhancements
Updated language data packs are available for Italian and French:
- fr-FR: 4.5.11
- fr-CA: 4.6.2
- it-IT: 4.5.4
These updated data packs address some tokenization mismatch between speech and text input for these languages. French expressions such as “envoyez-moi” and “l’orange” are now consistently tokenized as envoyez
-moi
and l'
orange
respectively.
If your projects use the hosted engine pack, and support French or Italian, make sure to upgrade them to the new data packs, rebuild your NLU models and domain language models, and deploy your applications again.
For self-hosted deployments, whenever the runtime engines are updated, you must upgrade projects, rebuild models and redeploy applications accordingly.
As well, NRaaS language groups for Canada and the United Kingdom are now available. For more information, see Language groups.
Documentation update
The Get Started section of the Event logs documentation was removed. The information in this section was outdated and no longer relevant.
2024-03-11
Documentation update
The Neural TTS documentation has been updated to clarify the differences between this service, which calls the Microsoft speech-to-text engine, and TTSaaS, which uses Vocalizer for Enterprise as its backend engine. See Neural TTS Essentials.
A new SSML example was added to the Neural TTS docs, showing how to include prerecorded audio in synthesis requests using the <audio> tag. See Prerecorded audio.
The event log documentation was updated for several services (ASRaaS, NRaaS, Dialog NII, and NTTS) to provide better examples for the client data field. See Event logs: asr-recognitioninitmessage payload, nrc-dtmfrecognitioninit payload, dlg-nii-logs payload, and tts-callsummary payload for Neural TTS.
2024-02-20
Documentation update
Added more information about capitalization and punctuation in ASRaaS results. See RecognitionFlags and Capitalization and punctuation.
Updated event log information for ASRaaS. See ASRaaS payloads.
2024-01-31
Enhancement
NRaaS language groups for Australia, China, India, Japan, Korea, and South America are now available. Also, the Portuguese language pack is now available in the Spain language group. For more information, see Language groups.
2024-01-24
Enhancement
In the DLGaaS API, a new field, language
has been added to the QAAction message in response payloads. This field indicates the language in which the dialog expects to collect user input.
See QAAction.
2023-12-21
Enhancement
NRaaS language group for Russia is now available. For more information, see Language groups.
2023-12-13
Enhancement
NRaaS language groups for France and Germany are now available. For more information, see Language groups.
2023-12-04
Enhancement
NRaaS now uses language groups to provide languages for recognitions. Language groups are groupings of language packs available in an instance of NRaaS. The URL you use to invoke NRaaS services determines both the region where recognition takes place and the language group used for recognition.
Documentation update
Adding information on new Date header in DLGaaS backend data access requests.
2023-11-14
Documentation update
New Responsible AI section, containing frequently asked questions about AI in Neural TTS.
2023-11-08
Enhancement
The following engine packs were released:
- Engine pack 2.4 for Speech Suite deployments
- Engine pack 3.11 for self-hosted Mix deployments
These engine packs support version 1.5 of the Dialog service.
For more information, see Manage engine packs and data packs.
Documentation update
Addition of NTTS payload for Neural TTS in the Event logs section.
2023-10-25
Enhancement
The TTSaaS documentation includes a new Python synthesis client. This sample client is the same as the one provided with Neural TTSaaS, but with different input files and instructions for each product. This means you can use the same sample client with either standard/enhanced or neural voices. See Sample synthesis client.
The HTTP Synthesizer documentation includes instructions for running it on Windows as well as Linux platforms.
Documentation updates
In Mix.nlu, added content clarifying how to handle sensitive personal information and problematic sample content in:
- Build your training set
- Add and import
- Apply automation
- Import Discover data
- Overview of entities
- Handling sensitive information
For DLGaaS, added some content clarifying the configurable session idle timeout. See Session idle timeout.
2023-09-28
General availability
Nuance Recognizer as a Service (NRaaS) is released for general availability and documentation is available here: Nuance Recognizer gRPC API. NRaaS is a speech recognition service for grammar-based, constrained vocabularies that allow natural speech in response to open-ended questions.
Currently, NRaaS is available for the United States geography only.
Documentation updates
To support the NRaaS general availability release, the following documentation additions were made:
- NR payloads in the Event logs section.
- About runtime engines in the APIs section to explain the differences between the runtime engines available in Mix.
2023-09-27
Documentation
Updates to DLGaaS:
- Added to Handling events reference topic to describe handling Too much speech recognition timeout error from ASRaaS.
- Updates to QAAction mask field.
2023-09-07
Enhancement
The webhook payloads for the available events now include additional information, such as the name of the user who requested (or approved) the action and the timestamp. See Webhook events.
2023-08-16
Documentation
In DLGaaS, some updates to reference topic content under Response actions and Data exchange and resources to improve clarity.
2023-08-02
Enhancement
The TTSaaS documentation includes instructions on running clients on Windows as well as Linux. See Sample client applications: Sample synthesis client and Sample storage client.
2023-07-26
Enhancement
A new German language data pack, NDP-DEU-DEU-GEN-4.5.1, is available. When upgrading from previous versions, you must rebuild any domain language models (DLMs) with the new data pack. See Building DLMs and NLU models.
2023-07-19
Enhancement
The ASRaaS documentation includes instructions on running clients on Windows as well as Linux. See Sample client applications: Sample recognition client, Sample training client, and Sample ForgetMe client.
2023-07-18
Feature highlight
Neural TTSaaS is released for general availability and documentation is available here: Nuance Neural TTS service. This new text-to-speech service uses Microsoft neural voices for natural-sounding and realistic speech in many languages. Neural TTSaaS offers 400 neural voices in 140 languages and accepts plain text and SSML input.
The Neural TTS service is available only for Nuance-hosted Mix projects. It is not supported with Speech Suite and is not available as a self-hosted engine.
In addition to the Nuance Neural TTS service documentation, new information has been added in the following sections of the Mix documentation for this feature:
- Mix.dialog: Configure TTS settings
- DLGaaS: Voice type and orchestration
- Mix.api: Voices
Mix.api release: 0.13.0
- Added the Voices endpoint.
2023-06-28
Enhancements
(External resources) It is now possible to define base URLs for external grammar references, including environment-specific overrides for environments in your deployment flow.
Notes:
- IVR applications using Nuance Speech Suite with VoiceXML Connector require engine pack 2.4 (or later) to support this feature.
- Applications in self-hosted environments don’t support this feature.
Mix.api release: 0.13.0
- Added a new environment configurations resource to support base URLs for external grammar references.
2023-06-21
Enhancement
(IVR applications) It is now possible to specify a transfer action type (such as blind transfer or route request) and destination, as such, when configuring external actions nodes. These Escalation settings are returned in responses by the DLGaaS API for use by the IVR application and are also logged in the transfer-initiated event.
Note:
This feature requires engine pack 2.4 (or later) for self-hosted Speech Suite deployments and Nuance-hosted IVR applications.2023-06-07
Enhancements
(Modalities) We now distinguish between input and output modalities, and explicitly introduce Text and Voice as input modalities to allow for more granular control over user input. At runtime, the Dialog service returns the input modalities set for the active channel, as well as only those resource references relevant to those input modalities.
Question and answer nodes also support setting input modalities locally, for example, you can disable speech input at a node that collects sensitive information to force a caller to use the keypad to provide the information instead of saying it out loud.
Input modalities set in question and answer nodes are now returned in the input-required event.
Notes:
- For self-hosted environments, this feature requires engine pack 2.4 (or later) for Speech Suite deployments; engine pack 3.11 (or later) for self-hosted Mix deployments.
- The first time you open existing projects in Mix.dialog, the two new input modalities, Voice and Text, will appear automatically, and both will be enabled by default.
- For hosted Mix projects, you can use the Targets tab of the Mix dashboard to disable any unwanted modalities in your channels. Once you’ve built and deployed your applications again, the Dialog service will be able to filter the resource references based on the input modalities you specified for the active channel.
Mix.api release: 0.11.0
- Added a new endpoint to update an application configuration.
- When starting a build you can now specify prebuilt domains (that is, NLU data sources).
- The mix.api.Modality schema now distinguishes between input and output modalities.
- Minor bug fixes.
2023-05-17
Enhancement
Dialog service now supports mTLS with client X.509 certificates to give additional security for server-side data transfers. This allows you to verify that a backend data request actually comes from the expected source.
For more details, see Dialog service mTLS support.
2023-05-09
Feature highlight
Copilot in Mix is officially in public preview. Copilot in Mix uses Microsoft’s Azure OpenAI Service to add GPT-powered features, giving you the ability to quickly try out ideas to improve customer engagement. For information on creating projects using Builder and Answers, both Copilot features, see Copilot in Mix (preview).
Note:
This feature is a public preview. Use of the feature is limited to testing and evaluation purposes only, with no production traffic.2023-05-04
Documentation
(DLGaaS) Added to Handling events reference topic to describe handling UserDisconnect events.
Fixed issue
(Mix.nlu) Previously regex collection type entities only worked for text input. NLUaaS can now match regex collection type entities in some cases to ASRaaS result input. This works when the ASRaaS formatted result is able to render the result into typical written form, for common patterns like dates, times, and phone numbers.
Note:
This currently works only with hosted NLUaaS and not with the Engine packs for Speech Suite or self-hosted Mix engines. Performance will depend on the specific language data pack installed.2023-04-28
Enhancement
(TRSX import) Importing a TRSX file that includes the NO_INTENT intent now returns a warning.
2023-04-26
Enhancement
(Event Logs) Credit card numbers that are between 13 and 19 characters and pass the Luhn algorithm test are now always masked in the Mix event logs. See Credit card numbers for more information.
Documentation
Added Known issues.
2023-04-24
Enhancement
(TRSX import) Previously, trying to import an intent or entity that already existed in the ontology returned an error. To improve usability, this operation now returns warnings. See View report for more information about errors and warnings.
Notes:
-
This behavior does not yet apply to the NO_INTENT intent; if the NO_INTENT intent is included in the TRSX file that is imported, an error message will be returned.
-
In multi-language projects, if an intent or entity already exists in the ontology for a language, importing a TRSX file with this intent or entity in another language will return warnings (“node name already exists”), even if the intent or entity does not yet exist in that language.
2023-04-20
Documentation
Added Integrating with Nuance Digital Engagement Platform. This new section gives tips for developers of digital customer service solutions.
2023-04-19
Enhancements
(Mix.dialog) New predefined variable: For multichannel applications, you can now use the predefined variable channel to switch to a different channel at runtime.
(Event logs) Two new events are now returned: channel-update indicates that the channel was updated for a transaction and language-update indicates that the language was updated for a transaction.
2023-04-12
Enhancement
(Mix.dialog) For complex variables based on the DynamicMessageReference schema, it is now possible to mark them as sensitive.
Fixed issue
(Dialog service) When the Dialog service evaluates an assignment at runtime, if the value to be assigned evaluates to null, the assignment is ignored, and the variable or entity on the left-hand side of the assignment remains unchanged, which might have unexpected outcomes.
Warning:
With this fix, the variable or entity is updated even when the value to be assigned evaluates to null. Since using a null variable or entity in a message will generate a null pointer exception, please consider reviewing your dialog designs to make sure you didn’t overlook the required logic to handle cases where data expected from a backend might be null (see Set conditions for more information).2023-04-05
Enhancements
The following engine pack was released:
- Engine pack 2.3 for Speech Suite deployments
This engine pack supports version 1.4 of the Dialog service.
For more information, see Manage engine packs and data packs.
ASRaaS introduces a new recognition flag, send_multiple_start_of_speech, that returns start-of-speech timing information for all utterances. Previously (and now by default), this information was sent only for the first utterance in the audio stream.
Mix is now available in the Canada geography.
2023-03-30
Fixed issue
The Get job details endpoint did not return errors that occurred when running the following Mix.api jobs:
This behavior gave the impression that a job was successful even when it was not. This issue has been fixed and error messages are now included in the report, providing accurate results.
2023-03-29
Enhancements
Mix.api release: 0.9.74
- Minor bug fixes
2023-03-22
Documentation
Expanding on our commitment to improve the documentation experience, we’ve replaced our search engine with Algolia. This is the first release—stay tuned for improvements in coming weeks.
2023-03-15
Documentation
Documentation was added for troubleshooting error code 119 - Topic not found when using the Mix event log API.
Enhancements
A new ForgetMe API has been added for the Mix runtime event logs. This API requests the removal of customer data in event logs stored outside the Nuance event log repository. See Forgetme gRPC API.
Mix.api release: 0.9.69
- Minor bug fixes
2023-03-08
Enhancement
(Mix.dialog) The terminating character for DTMF input (default: #
) is an optional setting: you can now clear this setting and leave it blank. See DTMF settings for more information.
2023-03-01
Enhancement
In DLGaaS, update to allow sending user input to DLGaaS in ASRaaS result format. See handling user input externally for more details. Also, update to send additional references to Mix resources (ASR, NLU, TTS, NR) in the response. These resources can be referenced by the client app for the purpose of carrying out client side orchestration.
2023-02-20
Enhancement
A new Public Preview version of Nuance Recognizer as a Service is now available. This version includes performance and reliability improvements.
Documentation
New topics have been added to the Nuance Recognizer as a Service documentation. These topics provide additional reference information for using the Nuance Recognizer gRPC API.
2023-02-09
Documentation
Documentation has been added for ASR models and resources best practices. The new section gives tips for creating domain language models (DLMs) and for leveraging resources to tune speech recognition.
2023-02-02
Enhancement
The following engine pack was released:
- Engine pack 3.10 for self-hosted Mix deployments
This engine pack supports version 1.3 of the Dialog service.
For more information, see Manage engine packs and data packs.
Fixed issue
(Dialog service) Sometimes, entities can be filled even in case of a nomatch event in question and answer nodes, which might have unexpected outcomes.
Note: The new default entity filling behavior in case of nomatch doesn’t automatically apply to existing projects, but it’s possible to migrate them manually, to benefit from this fix.
Warning:
For self-hosted environments, this feature requires engine pack 2.3 (or later) for Speech Suite deployments; engine pack 3.11 (or later) for self-hosted Mix deployments.Documentation
In NLUaaS, updates to Interpretation results: Intents to clarify the meaning of NO_INTENT and NO_MATCH in results. Also, some updates to Interpretation results: Entities to clarify how regex, rule-based, and predefined entities appear in results.
In DLGaaS, restructuring of the Handling inputs and outputs section. Addition of a new section on handling events.
The documentation for Nuance Transcription Engine (NTEaaS) has been removed, as the beta service was decommissioned.
2023-01-26
Enhancement
Hosted Mix projects have been migrated to a newer NLU engine version and the QuickNLP 10.2.x data pack series, providing:
- Improved accuracy for samples with audio input
- Enhanced support for predefined entities
For more general information on QuickNLP (QNLP) data packs, see Data packs.
2022-12-08
Enhancements
The Mix.nlu Discover tab provides a more granular time selector to query data. See Access and view data.
A new predefined variable, Last Message Object, holds the messages from the last visited message node or from the latest step (collection, confirmation, recovery, system actions) of the last visited question and answer node. It is only available for use as a Send data parameter in data access nodes, external actions nodes, and question and answer nodes. Note that the value of this variable is always masked in dialog event logs.
Warning:
For self-hosted environments, this feature requires engine pack 2.3 (or later) for Speech Suite deployments; engine pack 3.11 (or later) for self-hosted Mix deployments.ASRaaS added support for wordset weight, meaning you may assign a custom weight to wordsets within the resource stack, along with DLMs and builtins. In previous versions, wordsets had a fixed weight that could not be changed. The default wordset weight remains unchanged, so your applications will continue to run as before. See Resource weights.
Documentation updates
Documentation has been added for Nuance Recognizer as a Service. NRaaS understands and interprets speech and touchtone input based on the Nuance Recognizer engine, which uses grammars and/or trained language models.
(Mix.dialog) Added Entity lifecycle considerations.
2022-11-23
Documentation updates
(Mix.dialog) Updated Configure intent switching to clarify behavior when intent switching is disabled.
(Mix.dialog) Updated description of language, in Predefined variables, to mention TTS language-switching limitation and workaround.
(Integrating with Speech Suite) Added guidance against using separate Mix projects to design large applications.
(Mix.nlu) Added details about search bar, filter, and entity literals upload functionality in Entity panel.
2022-11-21
Enhancement
The following engine pack was released:
- Engine pack 2.2 for Speech Suite deployments
This engine pack supports version 1.3 of the Dialog service.
For more information, see Manage engine packs and data packs.
2022-11-17
Announcing the official launch of docs.nuance.com/mix/. Our redesign reflects your feedback and includes enhancements to navigation and a look and feel that reflects Mix design elements. Stay tuned as we continue to update links from within the Mix UI, work on improving search results, and add new content.
2022-11-16
Enhancements
(Mix dashboard) The All Organizations option has moved to the bottom of the Organizations list in the main Projects view. This change will prevent users from clicking the option accidentally and having to wait for the list of projects to load.
In addition, various enhancements were made to Organizations Users tab in the Manage tab. For details see Manage organization users. Note the two new roles available: Promotion approver and Pro services (PS), described in Organization roles.
2022-10-26
Feature highlight
After logging in to Mix you’ll land on the newly designed Mix Home tab. The Home tab provides new and experienced users a space to learn more about Mix, easily create new or access recent projects, see the latest announcements, join the Mix community, and more. For details, see Home tab.
Enhancements
A new Expert role is available, to provide users with access to more advanced features, such as the ability to upload rule-based grammars. See Organization roles.
Mix.dialog now provides support for emojis, symbols, and square brackets in messages. See Set messages for details.
Note:
To use emojis and symbols in messages or assignments, enter their graphic form directly into the text field. Unicode escape strings you might have in existing projects are supported for backward compatibility only. HTML decimal and hexadecimal equivalents are not supported.Mix.api release: 0.9.59
- The following Mix.api endpoints are now available to users with the Expert organization role:
- Added filter to Get list of Mix applications endpoint.
Fixed issue
(Event logs) The startTime
and durationMs
values currently returned in the Data processingTime
field cannot be used reliably.
Note: Nuance is evaluating the impact on reported latency in the event logs. Note that true latency has not changed—only the values reported in the event logs (startTime
and durationMs
) were incorrect.
2022-10-19
Feature highlight
Mix.dialog offers new functions that return the current date and time for the specified time zone. After storing the result of either function in a variable, you can use it in conditions and messages, in your dialog design.
Warning:
This feature requires engine pack 2.2 (or later) for Speech Suite deployments; engine pack 3.10 (or later) for self-hosted Mix deployments.Enhancement
The maximum value for the session idle timeout, set with the session_timeout_sec
field of the StartRequest message, has been increased from 90000s (25 hours) to 259200s (72 hours). The default session idle timeout remains at 900s (15 minutes). See Session lifetime in the DLGaaS documentation for more information.
Known issue
(Event logs) The startTime
and durationMs
values currently returned in the Data processingTime
field cannot be used reliably.
Fixed issue
(Mix.dialog) After deleting intents, stray data might remain in your dialog design, which generates warnings for intent mappers nodes when you validate your design. For example:
- Error finding intent with id 100d0595-c1c7-4138-98dd-7605afd40dac
- Found unexpectedly null intentions
Note: Once you have validated your design, Mix.dialog now lets you remove the stray data.
2022-10-05
Enhancement
Mix.api release: 0.9.56
- The mix.api.Organization message now includes whether deep learning model build, engine pack selection, or data pack selection is enabled for an organization.
2022-09-28
Feature highlight
It is now possible to set data types for entities in both Mix.nlu and Mix.dialog. A data type specifies what type of content will be collected by an entity. What were previously known as entity types are now called collection methods. Collection methods specify how the entity data is collected.
Data types allow dialog designers to apply methods and formatting appropriate to the data type in conditions and messages. Each data type has a number of compatible collection methods. Changing the collection method to another compatible collection method will not impact existing conditions and message formatting.
Warning:
This feature requires engine pack 2.3 (or later) for Speech Suite deployments; engine pack 3.11 (or later) for self-hosted Mix deployments.For more information, see:
- Mix.nlu: Data types
- Mix.dialog: Entity data types, User input properties, Get started, Compatibility between variables and entities, and Output formatting options
- TRSX: NLU model specification, now at version 2.6.
Enhancements
Support for hyphens (-
) in intent and entity names. For more information, see Naming guidelines. Note that “Model” was removed from the list of reserved keywords.
Note:
Intent and entity names cannot start with hyphens.Mix.api release: 0.9.55
- The mix.api.NLUModelType values were renamed:
- FAST is deprecated and was renamed
LEGACY
- ACCURATE is deprecated and was renamed
DEEP_LEARNING
- FAST is deprecated and was renamed
Mix.api release: 0.9.54
- Added new endpoints for data types:
- GET /v4/data-types
- GET /v4/ontology/entity-types](#get-entity-types)
- Updated multiple endpoints and schemas to include data types.
- The Project object now returns the ID of the organization to which the project belongs.
- Added new job status type,
STARTED
.
2022-09-21
Documentation updates
(Mix.dialog) Added Protection against infinite loops to mention the maximum number of times allowed looping through a node before reaching a question and answer node.
Warning:
This feature requires engine pack 2.2 (or later) for Speech Suite deployments; engine pack 3.10 (or later) for self-hosted Mix deployments.2022-09-08
Enhancement
A new language-specific speech setting, available at the project, channel, and node levels, allows you to tune the weight of the ASR domain language model built from samples defined in Mix.nlu, and referenced at runtime, when the Dialog service orchestrates with the ASR service.
For more information, also see the following Mix.dialog topics: Configure the weight of the ASR domain language model and Configure node-level settings.
Warning:
This feature requires engine pack 2.2 (or later) for Speech Suite deployments; engine pack 3.10 (or later) for self-hosted Mix deployments.2022-09-01
Enhancement
Users can now delete builds in the Manage tab. For more information, see Delete a build.
2022-08-24
Enhancements
It is now easier for newly added project members to find a project by selecting the organization, rather than choosing from the All Organizations list. Users must be part of the organization to have access from Applications and Organizations in the Manage tab.
Improvements were made to the CSV log produced when there are training issues, to include warning information as well as error information, and clearer messages about the sources of any issues. See Train and test your model for details.
2022-08-17
Enhancement
In multilingual projects, you can define different interactive elements for each language. Note that it is no longer required for an entity value to exist in all languages to be allowed to configure an interactive element for this value.
2022-08-08
Enhancement
Mix now provides the ability to add, modify, and enable/disable channels after a project has been created. See Add a channel.
Feedback
Was this page helpful?
Glad to hear it! Please tell us how we can improve.
Sorry to hear that. Please tell us how we can improve.