Version 4.16

Prev Next

Q3 2026 (Jul-Aug)

Enhancements

  • Node IDs Now Visible in Node Configuration
    Each node's unique ID is now displayed directly within its configuration panel in Data Flow, making it faster to reference specific nodes when troubleshooting or communicating with support.

  • Direct Link from Hound Jobs to Filtered Reports
    Hound job results now include a direct link to the Data Flow Reports page, pre-filtered to show only the sessions relevant to that job, removing the need to manually search or filter afterward.

  • Audio Player Added to the Reports Page
    An in-page audio player has been added to the Data Flow Reports page, letting users listen to call recordings directly from the report view without downloading them first.

  • Static Masking for Permanent Data Redaction
    Building on Dynamic Masking, tenants can now enable Static Masking to permanently redact sensitive personal data at the database level via a background process, giving customers with strict compliance requirements a way to irreversibly remove personal information rather than only hiding it on-screen.

  • Templates Available When Adding a Custom Node
    The "Add New Custom Node" dialog now offers ready-made templates, making it faster to get started when building a new custom node instead of beginning from a blank editor.

  • Updated AI Usage Disclaimer
    The AI usage disclaimer shown in Core Web has been updated to reflect SESTEK Core branding, along with refreshed legal text and links, ensuring the disclaimer accurately reflects the current product and terms.

  • Filler-Word-Aware Barge-In Detection
    Data Flow can now ignore filler words when detecting barge-in, reducing false barge-in triggers caused by callers making brief sounds rather than genuinely interrupting.

  • Template Support for Hound Bulk Processing
    Hound's Bulk Processor now supports templates, making it easier to configure and reuse common bulk-processing setups.

  • Stereo Audio and Channel Switch on the Test Page
    The Data Flow test page now supports stereo audio with a toggle to switch between channels, making it easier to test and review two-channel call scenarios directly from the test screen.

Improvements

  • Turkish Character Support Extended to Custom Node String Values
    Following the recent fix for Turkish character input in the custom node editor, string-type parameter values within custom nodes now also correctly support Turkish characters, closing a remaining gap in localized text entry.

  • Redesigned Voice and Language Selector
    The voice and language selector dropdown in Core Web has been redesigned for clearer navigation and easier selection when configuring TTS voices.

  • Readable Language Names in the Voices Table
    The Voices table now displays language codes as readable, human-friendly names instead of raw codes, making it easier to identify the language of each voice at a glance.

Bug Fixes

  • Fixed WebSocket Connections Not Closing When Leaving the Test Screen
    Resolved an issue where the WebSocket connection remained open after leaving the Test screen without pressing Stop.

  • Fixed an Incorrectly Pre-Filled TTS Retention Value for Automatic Tenants
    Resolved an issue where the TTS retention value was pre-filled incorrectly for automatically created tenants.

  • Fixed Unintelligible Stereo Recordings on the Reports Page
    Resolved an issue where stereo call recordings were unintelligible when played back from the Reports page.

  • Fixed TextManip Configuration Not Saving in the SR Node
    Resolved an issue where TextManip configuration settings in the SR node were not being saved.

  • Fixed Default Parameters Not Saving
    Resolved an issue where default parameter values were not being saved correctly.

  • Fixed Masking Not Being Applied on the Data Flow Reports Page
    Resolved an issue where masking was not applied to sensitive data shown on the Data Flow Reports page.

  • Fixed the "Mixed" Model Missing from the Language-Compatible Model List
    Resolved an issue where the "Mixed" model did not appear in the list of models compatible with the selected language.

  • Fixed an Error Opening the "About" Popup on the Data Source Page
    Resolved an error that occurred when opening the "About" popup on the Data Source page.

  • Fixed Audio Results from Different Operations Sounding Identical
    Resolved an issue where audio playback for different operations within the same session sounded identical, even when the underlying results differed.

  • Fixed Errors from the Data Source Storage Provider
    Resolved errors related to the data source storage provider.

  • Fixed Errors in the Reports Page Audio Player
    Resolved errors affecting the audio player on the Reports page.

  • Fixed a TTS Licensing Error
    Resolved an error affecting TTS licensing.

  • Fixed License Consumption Failures Caused by a Timezone Database Error
    Resolved an issue where license consumption could fail due to an error in the timezone database.

  • Fixed Masking Errors Preventing Results from Loading on the Reports Page
    Resolved an issue where masking-related errors on the Data Flow Reports page could prevent results from loading at all.

  • Fixed a Recognition Error with the SestekSystemTestModel
    Resolved a recognition error occurring with the SestekSystemTestModel.

  • Fixed a Noise Gate Error in SR
    Resolved a noise gate error affecting SR.

  • Fixed SR WebSocket Pronunciation Not Working
    Resolved an issue where custom pronunciations were not being applied to SR requests made over WebSocket.

  • Fixed SR Mangal Component Errors
    Resolved errors affecting the Mangal component in SR.

Fixed a SAPI.dll Registration Failure in TTS
Resolved an issue where SAPI.dll could fail to register correctly for TTS.


Compiled from closed and resolved Data Flow / Core Web work items for sprint v4.16, including bug fixes. Internal-only engineering tasks were excluded.