November 28th, 2025

Batch Container

Real-Time Container

GPU Transcription Inference Container

GPU Translation Inference Container

14.8.0 - Containers

Version 14.8.0 is now available for Batch Container, Real-Time Container, GPU Transcription Inference Container and GPU Translation Inference Container.

New

CPU & GPU

Batch and Real-time transcription

  • New Tagalog language is now available. Supports code-switching between Filipino and English for bilingual speech

GPU

Batch and Real-time transcription

  • Speaker identification – Introduces the ability to label speakers using unique speaker identifiers. Refer to documentation here for details

Real-time transcription

  • Channel diarization – Enables perfect speaker separation when there is one speaker per channel. Refer to documentation here for details

  • Channel and Speaker diarization – Enables separation of multiple speakers per channel. Refer to documentation here for details

  • Force end of utterance – Enables the client to force finalise transcription at the end of speech for faster finals (200ms), ideal when using external VAD or turn detection models for voice agents. Refer to documentation here for details

Improvements

GPU & CPU

Real-time transcription

  • Speaker Diarization – Improved speaker change detection accuracy for long audio streams (1+ hours)

GPU

Batch and Real-time transcription

  • Enhanced Operating Point

    • New models for English with improved accuracy in transcribing initialisms

    • Medical domain-specific model for Dutch, English, Finnish, French, German, Spanish giving the highest accuracy for healthcare use cases. Refer to the documentation for more details.

      • Relative accuracy improvements (medical domain): Dutch (nl) - 70%, English (en) - 14%, Finnish (fi) - 40%, French (fr) - 51%, German (de) - 36%, Spanish (es) - 63%

    • New models for Bilingual Malay English, Dutch, Finnish, French, Spanish with improved transcription accuracy

      • Relative accuracy improvements (general domain): Bilingual Malay English (en_ms)- 7.8%, Dutch (nl) - 16%, Finnish (fi) - 20%, French (fr) - 3%, Spanish (es) - 4.5%

  • Standard Operating Point

    • Enables faster transcription and higher throughput, refer to the documentation for more details

    • New models released with notable accuracy uplifts for the below languages:

      • Relative accuracy improvements: Belarusian (be) - 5%, Bulgarian (bg) - 7%, Catalan (ca) - 29%, Danish (da) - 22%, Finnish (fi) - 17%, Galician (gl) - 8%, Hungarian (hu) - 15%, Indonesian (id) - 14%, Korean (ko) - 8%, Latvian (lv) - 9%, Lithuanian (lt) - 11%, Marathi (mr) - 16%, Norwegian (no) - 12%, Romanian (ro) - 6%, Slovenian (sl) - 9%, Swedish (sv) - 6%, Thai (th) - 14%, Turkish (tr) - 32%, Ukrainian (uk) - 6%, Urdu (ur) - 15%, Vietnamese (vi) - 66%, Welsh (cy) - 17%

CPU

Batch and Real-time transcription

  • New Dutch models (Enhanced Operating Point) give up to 20% relative accuracy improvement

Real-time transcription

  • Improved Speaker Diarization accuracy for Enhanced Operating Point

Security fixes

GPU & CPU

  • A Software Bill of Materials (SBOM) is available for download from the corresponding release page in our Support Portal

  • Patched libgstreamer to address CVE-2025-3887. However, scanners are likely to still report it as vulnerable due to the unchanged base version string