Release Notes 7.1.0

Release date: 15 April 2026

Previous release: Release Notes 7.0.0

Version 7.1.0 of Capacity Private Cloud builds on the stability and capabilities introduced in version 7.0.0, delivering advances in speech synthesis, post-processing language coverage, Neuron integration, and developer tooling, along with improvements to the Analysis and Deployment Portals. 

New Features

The following capabilities were introduced in this release:

• Basque/Euskara TTS voice (Izaro) — Basque/Euskara support was added to Neural TTS with a new female voice named Izaro. Use Neural TTS model version 8.x for this.
• Full ITN for Portuguese — Inverse Text Normalization for Portuguese now includes punctuation restoration, capitalization, and redaction of sensitive data such as numbers, names, and identifiers. This brings Portuguese to production-ready parity with other fully supported ITN languages.
• ITN redaction for German — Redaction capability has been added to the German ITN implementation, complementing existing punctuation and capitalization support.
• Neuron transcription flow — The Neuron service now participates in transcription interaction flows, enabling intent and slot detection to run concurrently with transcription. This removes the need for a separate Neuron post-processing step in applicable workflows.
• Grammar cache-size control — A new Helm chart configuration option allows administrators to set a maximum on-disk grammar cache size per node. This prevents storage exhaustion in environments with limited disk capacity. Note: this requires a manual update to your Helm values file.
• Archiving for TTS sessions and grammars — The archive service now captures TTS session data and the grammars used in individual interactions. Grammars are now visible within interaction detail views in the Analysis Portal.
• Redis streaming for large files (>4 MB) — New Redis streaming helpers handle data files that previously exceeded message size limits, improving reliability for batch transcription and analysis set operations.
• Session service availability checks — The session manager now validates that required dependent services (ITN, Neuron) are available before creating an interaction. If a dependency is unavailable, a clear error message is returned rather than a silent failure.
• Deployment Portal multilingual support — The Deployment Portal interface is now available in Spanish, French, German, and Portuguese, in addition to English.

Updates & Improvements

The following existing capabilities were improved or extended in this release:

• Analysis Portal UI redesign — The portal has been rebuilt with updated Capacity branding. New functionality includes transcription review, grammar interaction details, audio markers, translation views, and contextual help panels. Post-Processing Profiles UI has also been updated with consistent icons, buttons, tables, and page header actions.
• Portal UI consistency — Addressed inconsistencies in button styles, confirmation dialogs, tooltip styles, navigation rail icons, dashboard responsiveness, and modal styles across both the Admin and Deployment Portals.
• Improved session manager logging — Ambiguous and low-value log messages in the session manager were revised to improve troubleshooting clarity.
• Reporting API queue handling — RabbitMQ queue and exchange configuration in the Reporting API has been corrected to prevent connection issues under load.
• Reporting API error messages — Misleading error messages returned by the Reporting API were rewritten to accurately reflect the underlying condition.
• Grammar structured logging — The service name used in grammar service structured logging was incorrectly changed in 7.0.0; this has been corrected in 7.1.0.
• Grammar log level — Resolved an issue where the grammar service was not honoring the cluster log level setting at startup.
• File-store logging — Resolved an issue where the file-store service was ignoring the cluster log verbosity setting.

Bug Fixes

The following defects were resolved in this release:

• Grammar crash loop — A specific grammar pattern caused the grammar service to crash with no log output, potentially grammar pods for approximately five minutes until the RabbitMQ request timed out. The grammar service now returns a valid parse output for the affected pattern and added more robust handling for similar error cases.
• Grammar load failures — Resolved grammar load failures related to server location handling.
• Grammar service disconnections — Resolved an issue with grammar-based ASR interactions that required grammar pod restarts, relating to a RabbitMQ reconnection issue within the grammar service.
• ASR pod traffic distribution — Resolved uneven traffic distribution across ASR pods.
• SSML address interpretation (Megan voice) — Resolved issues where the say-as interpret-as="address" requests were not correctly handling state abbreviations, street type suffixes, directional prefixes, and 5+4 digit zip codes.
• M4A/MP4 Redis errors — Fixed an issue where certain M4A and MP4 audio files caused Redis errors during processing.
• AMD/CPA grammar labels — Fixed incorrect grammar labels being returned for AMD and CPA requests when not using built-in grammars.
• ASR FinalResultStatus — Fixed an issue where interim results were incorrectly reporting FINAL_RESULT_STATUS_TRANSCRIPTION_MATCH for plain ASR interactions that did not involve transcription.
• Archive timing — Resolved an issue where sessions and interactions were being archived before their configured delay period had elapsed.
• Language Identification — 48kHz audio — Fixed “Audio too short” errors for PCM and WAV audio at 48kHz sample rate.
• Language Identification — MP3 misidentification — Fixed English MP3 audio at 8kHz and 11kHz incorrectly returning Chinese (zh) as the detected language.
• Analysis Set Prometheus counters — Fixed Prometheus metrics that were not incrementing when new analysis sets were created.
• Non-English locale pod failures — Fixed an issue where non-English locale pods were failing in version 7.0 deployments.
• License pod CrashLoopBackOff — Resolved license pod crash-loop issues present in version 7.0 deployments caused by RabbitMQ connection handling.
• Security — The following notable CVEs were addressed in this version:
  • File-store: CVE-2025-68121 (critical severity)
  • Grammar service: CVE-2026-33186

Patch Releases

The following patch releases were issued against version 7.0 containers and are included in the 7.1.0 baseline. Customers running 7.0.x patch releases do not need to apply these separately, and are instead encouraged to upgrade to this version, which includes additional improvements.

Installation Notes

Before installing or upgrading to version 7.1.0, review the following requirements. For full step-by-step installation guidance, refer to the Setting Up a Deployment article and the relevant setup guides in the Installation section of this knowledge base.

Infrastructure Requirements

The following are the recommended dependencies when using this version of the product.

• Kubernetes: version 1.35 minimum
• PostgreSQL: 17.7
• MongoDB: 8.2.4
• Redis: 8.4
• RabbitMQ: 4.1.8
• NGINX: 1.14.4

See our External Dependency Support Matrix article for more information.

Concurrency Limits

To maintain stable performance at scale, customers are encouraged to utilize the following per-pod concurrency limits, however individual use-cases vary, so adjustments for each specific use-case may be necessary:

• session, vad, lumenvox-api pods: maximum 100 concurrent interactions per pod
• grammar pods: maximum 50 concurrent interactions per pod

Usage Recommendations

• TTS text input: recommended maximum 4 MB per request
• Transcription audio: a maximum of 120 minutes per session is supported

Helm Chart Setup

Run the following command to update your local Helm chart repository before installation:

helm repo update

See our public Helm Charts repository for the latest and previously released versions, along with more information.

For MRCP deployments: there is no Helm chart for MRCP. Use the provided Docker Compose file available in the mrcp-api repository. The MRCP service runs on its own Docker virtual machine which then communicates with the configured Kubernetes cluster.

Neural TTS Configuration

To configure Neural TTS in your values file, specify the legacyEnabled flag per language. Set to false to enable Neural TTS (recommended); set to true to use legacy TTS. Neural TTS voice models must be explicitly listed in the values file so they can be installed and available to the system.

ttsLanguages:
  - name: "en_us"
    legacyEnabled: false
    voices:
      - name: "jeff"
      - name: "aurora"

ITN Configuration

When deploying ITN, include the relevant language entries in your Helm values file:

itnLanguages:
  - name: "en"
  - name: "es"

Add additional language entries (e.g., pt for Portuguese, de for German) as required by your deployment.

Upgrade Procedures

Upgrading from previous versions of Capacity Private Cloud is supported. Contact Capacity Support to discuss your specific upgrade path before proceeding, particularly if upgrading from a version earlier than 7.0.

When upgrading, ensure your NGINX ingress version is updated to at least 1.14.4 before beginning the process.

If upgrading Neural TTS from version 6.0.0, the TTS cache folder and the Neural TTS models folder must both be cleared before starting the upgrade. Contact support if you have questions about this step.

API Reference

See our API documentation for 7.1.0 for details about this version's API

Model Versions

Various acoustic, language and other AI models are distributed for use with our services, and the versioning of those models does not following the same version as the corresponding software release, since the models may be modified less frequently. The following model versions were introduced and designed for this release:

• ASR — 7.x
  • High-Definition (HD) models available for: English, Spanish, French, Italian, Japanese, and Portuguese
• Neural TTS — 8.x
  • Sample rate: 16 kHz (can be downsampled to 8 kHz)
  • New voice: Izaro (female, Basque/Euskara)
  • Legacy TTS models continue to run under version 1.0
• ITN models updated: Portuguese (full support), German (redaction added)