Local console

Transcription Engine

Upload media, run transcription, manage hotwords, and download exports for review.

Proprietary research previewDiplo Transcription Engine is a product of DiploFoundation (diplomacy.edu) and is proprietary software running on an NVIDIA DGX Spark system located in the Belgrade office. This deployment is for evaluation, validation, and stress testing only; it is not production ready.All transcription processing, file storage, hotword management, transcript exports, and debug logs run locally on this server. Uploaded media and generated outputs are not sent to external APIs, cloud services, or third-party systems.The system demonstrates cutting-edge ASR research workflows and may produce incomplete, inaccurate, or unstable results during testing. Access should remain limited to approved testers on trusted local or forwarded connections.Uploaded media, hotwords, transcripts, exports, and debug logs stay on this machine under data/ and audit/ and may contain sensitive content. All transcripts are automated drafts and require human review before operational, legal, publication, or archival use.
1

Media file

Select the audio or video file to transcribe.

Upload

Drag a file here, or browse

MP4, MOV, WAV, and other FFmpeg-readable files are supported.

No media selected.
2

Session details

Use clear IDs so outputs are easy to find later.

3

Output profile

Pick a preset, then adjust files if needed.

Review default keeps normal operator settings.

4

Engine and hotwords

Confirm the backend and select terminology files.

Loading hotword files...

Advanced transcription controls
HW

Hotword Manager

Create terminology files and edit terms without touching JSON manually.

Terms

Use Edit to load a row into the form. Remove deletes it from the JSON file.