Guides
Guides
Breadcrumbs

July 2026

AGI-Connector 4.23.0

Improvements

  • Reduced False Error Alerts on Caller Hang-Up - When a caller ends a call normally, including short calls or platform-initiated disconnects, the connector no longer sends an error notification to the contact center platform. This eliminates false alert emails and error signals that were previously triggered for calls that ended without any issue.

  • IVR Error Code Visibility - When a call encounters a speech or prompt failure, the connector now surfaces the specific IVR error code (for example, a prompt playback failure or a grammar error) instead of reporting all failures under a single generic label. Omilia Support teams can identify the type of failure at a glance, reducing the time it takes to diagnose and resolve the issues.

Fixes

  • Reliable Transfer Completion for Genesys Calls - Calls transferred from OCP® to a Genesys agent now reach the agent consistently. Previously, a small percentage of calls were returned to OCP as if they were new, causing callers to repeat their interaction instead of being connected to the agent they were transferred to.

  • This release includes security updates.

AudioHub 7.2.1

Fixes

  • Prevented Critical Errors - Improved audio retrieval API response time and reliability by eliminating an unnecessary database query.

AudioHub 7.3.0

New Features

  • Speaking Rate and Volume Control for Google Chirp 3 HD Voices - You can now control speaking rate and volume when synthesizing speech with Google Chirp 3 HD voices.

Improvements

  • Expanded Language Support - AudioHub now supports an extended set of languages and locales, including Norwegian, Swedish, Japanese, Mandarin Chinese, Cantonese, Filipino, Malay, Thai, Indonesian, Hindi, and a range of regional Spanish, German, English, Dutch, Hebrew, Danish, and French variants. This broadens the reach of voice applications to more global audiences.

Chat-Connector 8.18.0

New Features

  • Automatic Language Detection for Chat Sessions - The chat widget now detects the visitor's browser language automatically and selects the matching conversation configuration, ensuring visitors are served in their preferred language without requiring separate API keys for each locale. If no matching language is found, the session falls back to the default locale seamlessly.

  • Chatter IP Address Visibility in Dialog Logs - The IP address of the person chatting is now captured and displayed in OCP Conversational Insights®, giving your team additional context for antifraud analysis and session identification.

  • Secure Sensitive Information Capture in WebChat - When a dialog requests sensitive information such as a password, the chat widget can now mask the input field and preserve the exact characters typed, including uppercase letters, ensuring credentials are handled securely throughout the conversation.

Improvements

  • Chat Input Locked During Agent Transfer - While a conversation is being transferred to a live agent, the chat input is disabled and displays a waiting message. Input is re-enabled automatically once the agent connects, preventing messages from being sent into an unattended session.

Fixes

  • Accurate Agent Join Detection for Salesforce Transfers - Supervisors monitoring a Salesforce chat session are no longer mistakenly treated as joined agents, so the handoff experience behaves as expected.

miniApps 2.42.1

Improvements

  • Base64 Encoding and Decoding in JavaScript Scripts - The `atob` and `btoa` functions are now available in all JavaScript scripts running in isolated execution mode, enabling base64 encoding and decoding without extra libraries or workarounds. Scripts that previously relied on these functions when running outside isolated mode now work consistently regardless of the execution strategy in use.

Fixes

  • Reliable Handling of Non-ASCII Characters in Script Requests - Scripts and parameters containing non-ASCII characters, such as accented letters, special symbols, or non-Latin text, are now decoded correctly even when the request body is received in multiple parts. Previously, these characters could be corrupted intermittently, causing unexpected script execution errors.

miniApps 2.43.0

New Features

  • Dutch Language Support for OCP miniApps® - OCP miniApps® now support Dutch (nl-NL), enabling contact centers serving Dutch-speaking customers to deliver fully localized conversational experiences out of the box.

  • Flemish Language Support for OCP miniApps® - OCP miniApps® now support Flemish Dutch (nl-BE), extending localization coverage for contact centers operating in Belgium and serving Flemish-speaking callers.

  • Confidence-Based Intent Clarification - A configurable confidence threshold can now be set per intent, allowing the system to automatically trigger clarification prompts when a caller's intent is not recognized with sufficient certainty. This reduces misrouted calls and improves the accuracy of automated conversations without requiring manual intervention.

STT 8.8.0

New Features

  • Improved Liveness Detection Accuracy - deepASR® now streams a brief period of silence before speech audio begins, giving the liveness detection engine more complete audio context. This reduces misclassification at the start of calls, providing more reliable detection of whether a live person or an answering machine has picked up.

Improvements

  • More Reliable Number Transcription for GENERIC ASR (logos models) - deepASR® now applies a post-processing step to Generic ASR transcription results, converting spoken numbers (zero through twelve, including "oh" for zero) into digits when the engine's built-in normalization does not. This reduces transcription errors in scenarios where callers speak numeric values such as zip codes, account numbers, or menu selections.

  • Faster and More Consistent Recognition Completion - Resolved a condition where simultaneous internal processing events could cause recognition to stall for up to two seconds before returning results. Speech recognition now completes promptly and consistently, regardless of the order in which audio processing events arrive.

Fixes

  • Reliable Session Log Uploads - Session log files were not being uploaded after early April 2026. This has been resolved by consolidating the upload service directly into deepASR®, ensuring session logs are captured and stored reliably for every call.

  • Prevented Engine Failures on Large Audio Frames - deepASR® now enforces a maximum size on incoming audio chunks before they are forwarded to the recognition engine. Previously, oversized audio frames could corrupt the internal audio stream, causing the recognition engine to fail for the remainder of the session and forcing a session restart.

  • Prevented Startup Crash When Voice Biometrics is Misconfigured - deepASR® no longer crashes on startup when the biometrics service is disabled but the voice verification service is also enabled. The service now detects the inconsistency and disables voice verification automatically, ensuring a clean startup in all configuration combinations.

  • Resolved Stuck Engine Incident Causing Call Failures - A production incident where all deepASR® recognition engines became occupied and unable to process new calls has been investigated and resolved. Improvements to thread management and internal timeout handling prevent engines from becoming permanently stuck, restoring normal call processing without requiring a restart.

  • This release includes security updates.

Testing Studio 1.20.0

New Features

  • Add support for testing concierge / router agents in Agents Unit Testing (AUT). Users can now define an expected route for a scenario, and the scenario UI adapts to the agent type so only the relevant fields are shown.

Improvements

  • Update LLM model for our LLM features.

  • Speedup Run tests clone task.

  • Enhance the Agents Unit Testing (AUT) CSV export/import so that exported results can be re-imported cleanly as scenarios, with full support for both QnA and Concierge agent types, route-based grading, and a safe import preview.

TTS 3.0.0

Beta

Lexis is Omilia’s new generative text-to-speech model, built to unlock the most authentic Customer Service experiences.

Speech with the expressiveness, warmth, timing, and emotion of a skilled human agent, generated in real time, entirely inside your own platform and adapting to the context and the stakes of every conversation, in many languages.

  • Model: Lexis v1.0.0

  • CX Ambassador Persona library: en-US - Rob v1.0.0

  • SSML support: <break>

New features

  • New voice family Lexis. Adds support for our newest, more natural-sounding voice technology.

  • New SSML <sub> tag. You can now control exactly how a word or abbreviation is spoken by supplying a spoken-form alias. Example: <sub alias="World Health Organization">WHO</sub> is read as "World Health Organization" while "WHO" stays in your text.

For more information go to SSML tags support of Omilia TTS.

xPacks 1.18.3

Fixed OCP xPacks

  • en-GB Universal deepNLU® xPack v 3.16.0

    • Improved recognition accuracy for ACCEPT, Agree, Alphanumeric, CardNumber, CONVENTIONAL, DIALOGACT, Exit, Hour, Numeric, REJECT, SentimentDescriptor, and YesNo entities.

    • The Numeric entity now returns all candidate interpretations of composite spoken numerals (e.g. "twenty nine" → 29, 20, 9) when the caller is interacting with Numeric or Alphanumeric miniApps.

    • The words "space" and "hyphen", when spoken during Alphanumeric dictation, are now filtered out. The resulting annotation remains continuous.

    • Phrases such as "it is", "it does" (ACCEPT) and "it is not", "it does not" (REJECT) are now reliably mapped to the correct dialog act entities. Previously these were missed or only partially triggered.

    • Several Hour entity misrecognition cases have been addressed, including false Hour annotations triggered by numeric ordinals (e.g. "one time", "two times") and incorrect value assignments for expressions like "all night and all morning" and "three o’clock this morning".

    • Improved recognition of card numbers provided as compound spoken numerals. Previously, such inputs were truncated to partial values due to incorrect numeral decomposition.

  • en-GB Universal+ deepNLU® xPack v 3.16.0

    • Improved recognition accuracy for ACCEPT, Agree, Alphanumeric, CardNumber, CONVENTIONAL, DIALOGACT, Exit, Hour, Numeric, REJECT, SentimentDescriptor, and YesNo entities.

    • The Numeric entity now returns all candidate interpretations of composite spoken numerals (e.g. "twenty nine" → 29, 20, 9) when the caller is interacting with Numeric or Alphanumeric miniApps.

    • The words "space" and "hyphen", when spoken during Alphanumeric dictation, are now filtered out. The resulting annotation remains continuous.

    • Phrases such as "it is", "it does" (ACCEPT) and "it is not", "it does not" (REJECT) are now reliably mapped to the correct dialog act entities. Previously these were missed or only partially triggered.

    • Several Hour entity misrecognition cases have been addressed, including false Hour annotations triggered by numeric ordinals (e.g. "one time", "two times") and incorrect value assignments for expressions like "all night and all morning" and "three o’clock this morning".

    • Improved recognition of card numbers provided as compound spoken numerals. Previously, such inputs were truncated to partial values due to incorrect numeral decomposition.

  • en-AU Universal deepNLU® xPack v 3.7.0

    • Improved recognition accuracy for ACCEPT, Agree, Alphanumeric, Amount, CardNumber, CONVENTIONAL, DIALOGACT, DayOfMonth, Exit, Hour, MonetaryAmount, Numeric, REJECT, SentimentDescriptor, TimeRelativeToNow, and YesNo entities.

    • The Numeric entity now returns all candidate interpretations of composite spoken numerals (e.g. "twenty nine" → 29, 20, 9) when the caller is interacting with Numeric or Alphanumeric miniApps.

    • The words "space" and "hyphen", when spoken during Alphanumeric dictation, are now filtered out. The resulting annotation remains continuous.

    • Phrases such as "it is", "it does" (ACCEPT) and "it is not", "it does not" (REJECT) are now reliably mapped to the correct dialog act entities. Previously these were missed or only partially triggered.

    • Several Hour entity misrecognition cases have been addressed, including false Hour annotations triggered by numeric ordinals (e.g. "one time", "two times") and incorrect value assignments for expressions like "all night and all morning" and "three o’clock this morning".

    • Improved recognition of card numbers provided as compound spoken numerals. Previously, such inputs were truncated to partial values due to incorrect numeral decomposition.


Webchat 2.37.0

New Features

  • Chat Transcript Download During Active Conversations - The "Download as PDF" button in OCP Conversational Chat® is now available throughout an active conversation, not only after it ends. Users can download a full transcript of all messages exchanged so far at any point during the session, enabling mid-conversation reference and record-keeping.

  • Sensitive Input Masking for Secure Data Entry - When a conversation requires confidential information, the chat widget can now mask each character as it is typed, replacing it with asterisks. An eye icon lets the user reveal what they typed before sending, and the chat history displays the value as masked, protecting sensitive data from being visible on screen.

  • Input Disabled During Agent Transfer - While a conversation is being transferred to a live agent, the chat input field and send button are automatically disabled and a waiting message is shown. Input re-enables as soon as the agent joins, preventing messages from being sent into an unmonitored session during handover.

Fixes

  • Accessible Time Picker via Keyboard Navigation - Users navigating the chat widget by keyboard can now fully select and confirm a time in the time picker without requiring a mouse.

  • Localized Interface Text for Updated Chat Header - Text fields introduced in the updated chat header are now correctly translated across all supported languages, ensuring a consistent experience for all users.