August, 2026

Introduction

This month brings updates across the Sanas App, Sanas Chrome Extension, Language Translation, and Speech Intelligence. The Sanas App adds in-app warnings, taskbar notifications, a network checker, and Linux readiness improvements, while the Chrome Extension now supports Language Translation, which itself gets several upgrades around utterance completion, connection handling, and audio settings. Rounding things out, Speech Intelligence introduces a Custom Glossary, Test Datasets, an Alert Library, and improved self-serve tools for agents and admins.

What’s New

• In-App Warnings and Error Notifications now surface key issues directly in the app, helping users identify and resolve problems before they impact an active call.

• Taskbar Notifications now provide an at-a-glance indication of the Sanas App’s call readiness and status without requiring users to open the app.

• In-App Network Checker now displays network status and Sanas URL reachability, helping users quickly identify connectivity issues on their side.

• Sanas App x Linux Readiness now supports SSO and Gmail login, along with installer IDs to enable build traceability and support self-serve readiness on Linux.

• Sanas Chrome Extension now supports the following:

  • Language Translation

  • Speech Enhancement

  • Inbound Accent Translation

  • Remote Inferencing

• Language Translation Upgrades and Fixes

  • Utterance card now turns green once the full translated speech has played, indicating it’s safe to place the caller on hold.

  • Connection failure handling now transitions to a connection failure state when an attempt fails, instead of remaining stuck in connecting mode.

  • Translated Speech setting now lets users disable incoming translated speech audio, reducing wait time and improving round-trip latency while relying on the in-app transcript

• Speech Intelligence Enhancements

  • Custom Glossary is a new feature for managing customer-specific terms to improve transcription accuracy.

  • Test Datasets is a new library under Transcript Management for building and managing reusable test transcripts to validate Speech Intelligence configurations.

  • Agent Portal Reliability & Parity brings the Agent Portal onto live SCI data with fixed alert windows and metrics now consistent with the Supervisor Portal.

  • Platform Improvement adds URL-based tab persistence to Speech Intelligence settings, enabling shareable deep links and browser back/forward navigation.

  • Self-Serve Agent Portal for agents to view their own analytics, call history, call details, and compliance alerts without supervisor-wide group access.

  • Alert Library has been added in the Sanas Portal, allowing Account Admins to browse available alert rules and enable or disable them for each group without requiring code changes for routine updates.

  • Unique Alerts & Stats improvements for Compliance Alerts add optional per-rule cooldowns to prevent duplicate in-call alerts and switch hit-rate calculations to distinct calls for more accurate reporting.

  • Call Boundary Detection now uses context-based signals and silence heuristics to detect call starts and ends without relying on microphone engagement signals.

  • Live Tracking now provides server-backed, stable sorting for Name, Alerts, Duration, and Sentiment across the entire active roster, with improved real-time pagination and resolved roster flickering during sort changes.

  • Compliance Alerts now support searching by User ID with ClickHouse-powered filtering, enabling faster agent lookup and improved performance for large groups.

  • Transcript Transfer now includes Amazon S3 connection testing with required validation before saving, providing a more reliable and guided setup experience.

In-App Warnings and Error Notifications

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

The app now surfaces warnings and errors directly in the interface, helping users identify and address issues before they impact an active call. Currently handled cases include:

• Non-USB device selected

• Model load errors

• Audio device failures

• High CPU usage

Note: Configuration in the Sanas Portal is not required. This is available to all app users by default.

Taskbar Notifications of the Sanas App

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

The badge on the Sanas icon in the taskbar now provides an at-a-glance indication of call readiness, so users can check the app’s status without opening it:

• Red — Error state

• Amber — Warning state

• Green — Correctly set up and in a call

• No badge — Correctly set up and not in a call

Note: Configuration in the Sanas Portal is not required. This is available to all app users by default.

In-App Network Checker

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

A new in-app utility displays the current network status and Sanas URL reachability, making it easier to determine whether a connectivity issue is on the user’s side.

Note: This in-app utility is available on the login page and, after signing in, under More in the bottom-left corner of the Sanas app. Configuration in the Sanas Portal is not required. This is available to all app users by default.

Sanas App x Linux Readiness

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

• Linux — SSO and Gmail Login: Users on Linux can now sign in using SSO or their Gmail account, bringing the Linux login experience in line with other platforms.

• Linux — Installer ID Support: The Linux installer now carries an installer ID, allowing builds to be tied back to their distribution source. This is a prerequisite for self-serve readiness on Linux.

Language Translation in the Sanas Chrome Extension

Sanas Chrome Extension version: 1.9.1 and later

Upon logging in to the Sanas Chrome Extension, a new icon gets displayed — Language Translation toggle. By default, the toggle is OFF.

To utilize this product, follow the these steps:

1. Click the Language Translation icon.

   

2. Turn the toggle ON.

   

3. To display the captions, turn the Show captions toggle ON.

   

Below is a sample image of an active and working Language Translation on Chrome Extension:

By default, Language Translation starts automatically on your browser-based softphone or conferencing app once it is set up and ready to use.

Speech Enhancement in the Sanas Chrome Extension

Sanas Chrome Extension version: 1.10.3 and later

Speech Enhancement is now available in Chrome Extension. To leverage this:

1. Click the Accent Translation icon upon logging in to the Chrome Extension.

   

2. Select Speech Enhancement tab to switch to this product.

3. Enable the What you say and What you hear toggles to fully utilize Speech Enhancement during your calls.

   

   Note: What you say enhances the way you speak (outgoing audio), while What you hear boosts the way your recipient speaks (incoming audio).

Inbound Accent Translation in the Sanas Chrome Extension

Sanas Chrome Extension version: 1.10.3 and later

Accent Translation for Incoming Audio is now available in Chrome Extension. To utilize this:

1. Click the Accent Translation icon upon logging in to the Chrome Extension. By default, the toggle is OFF.

   

2. What you hear toggle is for Inbound Accent Translation, while What you say toggle is for Outbound Accent Translation. Turn both of them ON to full utilize the product.

   

Remote Inferencing in the Sanas Chrome Extension

Sanas Chrome Extension version: 1.10.3 and later

Available for both Accent Translation and Speech Enhancement, here is how to enable Remote Inferencing in the Sanas Portal for Chrome Extension:

1. In the Sanas Portal, click on Settings, then Configuration.

   

   Note: This toggle is typically accessible to Sanas Admins only. Please contact our Support team to enable this feature before proceeding to the next steps.

2. SIP endpoints can be set, depending on the type of product your organization subscribed into.

   

3. Under the Products section (Settings > Configuration > Products), look for Inferencing Mode, then set your selection:

   On Device or Remote Inferencing

   

   Note: Setup is similar for Accent Translation and Speech Enhancement.

Utterance Speech Completion for Language Translation

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

Once the agent’s translated speech has finished playing to the caller, the utterance card turns green, clearly indicating that the caller has heard the full translation and it’s safe to place them on hold.

Note: This is a fix applied to the Sanas App version 5.8.x and is not tied to any specific Language Translation models.

Connection Bug Fix for Language Translation

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

Language Translation now transitions to a connection failure state when a connection attempt fails, instead of remaining stuck in connecting mode.

Note: This is a fix applied to the Sanas App version 5.8.x and is not tied to any specific Language Translation models.

Disable Incoming Translated Speech Option for Language Translation

Sanas Desktop App version: 5.8.5 (Self-Serve) and 5.8.6 (Direct-CX), and later

There is now a Sanas Portal setting to disable translated speech audio from the caller, allowing users to rely on the in-app transcript instead. This can reduce the time spent waiting for incoming audio and improve overall round-trip latency.

• How it works?

  From the Sanas Portal → Settings → Configuration → Language Translation

  

  

  Note: This shows up in the Sanas application with an indicator for Translated Speech, as shown above.

Custom Glossary for Speech Intelligence

The Custom Glossary feature of Speech Intelligence lets you teach the transcription engine customer-specific vocabulary — brand names, people, places, and domain jargon — so it recognizes and transcribes these terms correctly instead of mishearing or generalizing them.

How does it work?

1. From the Sanas Portal, account admins can utilize this feature by following this path:

   Sanas Portal > Settings > Products > Speech Intelligence > Custom Glossary

2. To start, enable the Use Custom Glossary toggle ON.

3. Add the glossary terms one by one or bulk upload them (up to 1000 terms).

Dataset Testing for Speech Intelligence

Test Datasets provides a library of reusable, PII-redacted transcript datasets for testing Speech Intelligence configurations before they go live.

How does it work?

1. From the Sanas Portal, account admins can utilize this feature by following this path:

   Sanas Portal > Settings > Products > Speech Intelligence > Transcript Management > Test Datasets

2. Account admins can perform the following in this section:

   1. Dataset Library / Dataset Scripts views – Toggle between browsing saved datasets and managing the underlying dataset scripts.

   2. Search & filter – Search by name or description, and filter by source to quickly locate the right dataset.

      Upload Dataset – Bring in your own custom transcripts for testing.

   3. Sample from Production – Pull real, PII-redacted transcript samples directly from production calls, scoped to your hierarchy, to build realistic test data.

   4. Generate Synthetic – Create synthetic test transcripts when real samples aren't available (Coming soon!)

   5. Copy/edit support – Datasets can be duplicated and modified for iterative testing.

Update: Synthetic Data Generation is now available! The Synthetic Dataset Wizard supports multiple seed sources—write a prompt, select a production call transcript, or upload a pre-redacted seed transcript—then generates variant transcripts for testing configurations before go-live.

Agent Portal Reliability and Parity for Speech Intelligence

Agent Portal now runs entirely on live SCI data in production, with the fixture fallback removed. The alert date window now passes correctly, and call duration, discussion points, and timestamps are now consistent with what's shown in the Supervisor Portal.

Platform Improvement for Speech Intelligence

The Speech Intelligence settings tabs — including the nested Transcript Management tabs — now persist the selected tab in the URL, enabling shareable deep links and proper back/forward browser navigation.

Sample instance:

• For the Configuration section of the Speech Intelligence panel under the Settings page:

  

Self-Serve Agent Portal for Speech Intelligence

Agents can now access the Speech Intelligence dashboard without supervisor-wide group access, giving them visibility into their own SI analytics, Call History, Call Details (including transcripts), and Compliance Alerts.

Alert Library for Speech Intelligence

Speech Intelligence Configurations now includes an Alert Library (Production), allowing Account Admins to browse available alert rules and enable or disable them for each group directly in the Portal—without requiring code changes for routine rule updates.

Update: Speech Intelligence Configuration expands the Alert Library into a full authoring workspace: Production, Non-production, and Archive shelves, with version-controlled alert rules per group (create/edit versions, promote, archive, and restore). Per-alert configuration (enable/disable and related rule settings) is stored as group-owned settings, so routine rule management no longer depends on code or one-off DB changes.

Context-based Call Boundary Detection for Speech Intelligence

Updated call tracking to rely on context-based signals and silence heuristics, removing dependency on microphone engagement signals.

• Call ends when:

  • SI model signals the end of the call

  • No speech is detected on the customer’s channel for 5 minutes

  • SI model signals the start of a new call

• Call starts when:

  • No call is active, but speech is detected on the customer’s channel

  • A call is active, but the SI model signals the start of a new call

Unique Alerts & Stats for Compliance Alerts

The Compliance Alerts for Speech Intelligence has been improved, and included the following:

• Unique Alert: Optional per-rule cooldown prevents the same in-call fraud alert from firing repeatedly on a single call within a configured window, cutting down on duplicate live noise.

• Hit Rate Accuracy: Compliance alert hit rate now counts distinct calls hit rather than raw firings, keeping rates at ≤100% and aligning with how supervisors interpret "how often did this fire on calls."

Enhanced Live Tracking Sorting Controls for Speech Intelligence

Speech Intelligence model version: 1.5.0, and above

• Live Tracking now supports server-backed sorting for Name, Alerts, Duration, and Sentiment, enabling supervisors to consistently sort the entire active roster across all pages instead of only the current page.

• Sorting and pagination have also been enhanced to maintain stable ordering for alerts, duration, and sentiment as the roster updates in real time. Additionally, an issue causing calls within the grace period to flicker when changing sort order has been resolved.

Note: This release goes together with new Sanas Portal version 1.19.0, and above.

Expanded Search Capability of Compliance Alerts for Speech Intelligence

Speech Intelligence model version: 1.5.0, and above

• Compliance Alerts statistics now support searching by User ID, enabling supervisors to quickly locate an individual agent's alert breakdown. Filtering is now processed directly in ClickHouse, significantly improving search performance for large groups.

Note: This release goes together with new Sanas Portal version 1.19.0, and above.

Improved Transcript Management for Speech Intelligence

Speech Intelligence model version: 1.5.0, and above

• Transcript Transfer settings now include a Test Connection flow for Amazon S3 destinations, allowing Speech Intelligence to verify write access before saving the configuration. The setup experience has also been improved with clearer guidance and validation that requires a successful connection test before the destination can be saved.

Note: This release goes together with new Sanas Portal version 1.19.0, and above.

Support