Product

Local AI transcription for sensitive environments

Windows desktop application. Fully offline. VS-NfD-ready. MDM-compatible. No byte leaves the workstation.

Get in touch

Capture mode

Control from the system tray

TALDEN runs as a system tray application. Three states, full control, no window required.

IDLE

Application active, no audio access. Waiting for hotkey.
RECORDING

Microphone active. Real-time processing on-device. Zero network connections.
PAUSED

Recording suspended. Session data remains local.

Processing pipeline

Three stages. Fully local.

Whisper transcribes the conversation. pyannote separates speakers. Mistral structures the minutes. All models run on-device.

Use-case profiles

Six predefined profiles

Each profile optimizes transcription, structure, and tone for a specific conversation type. Extensible on request.

Meeting

Free conversation → Structured minutes

**Summary:** Q3 planning kick-off. Budget approved.
**Decisions:** Launch target Aug 15 · Max owns frontend
**Open items:** Staging environment by Friday

Email

Spoken draft → Polished message

Dear Mr Hoffmann,
Thank you for yesterday's conversation. Please find my proposal for our continued collaboration...

Rage

Emotional rant → Professional message

I'd like to flag that the agreed deadline was moved without consultation. I'm requesting a firm new date by tomorrow.

Brainstorm

Stream of consciousness → Idea map

**Productivity:**
  · Async-first mode
  · Auto-agenda from minutes
**Sales:**
  · DACH direct sales Q4

Interview

Recorded conversation → Clean Q&A

**Q:** How do you handle data privacy for recordings?
**A:** All processing is local. No byte leaves the device.

Notes

Spoken notes → Structured text

· Meeting legal: Mon 10am — review NDA draft
· Server migration: backup first
· Q4 budget: approval pending

System requirements

Workstation requirements

Component	Minimum	Recommended
Operating system	Windows 10 version 2004+	Windows 11
Memory	8 GB RAM	16 GB RAM
Processor	Intel Core i5 10th gen.	Intel Core i7-1185G7+
GPU	Integrated graphics (CPU mode)	NVIDIA RTX 3060 / Quadro T500
Storage	25 GB free space	NVMe SSD
Network	No connection required	Firewall isolation recommended

Deployment

Installation in 7 steps

MSI package for Windows. Silent install and MDM deployment fully supported.

01

Obtain installer

MSI package delivered via secure, verified channel.
02

Verify signature

Verify Authenticode signature (SHA-256) before installation.
03

Silent install

msiexec /i TALDEN.msi /quiet — no user interaction required.
04

Deploy models

Place AI model package in %PROGRAMDATA%\TALDEN\models\.
05

Configure firewall

Block outbound connections. TALDEN requires no network access.
06

Set MDM policy

Configure via Registry key or GPO. Full documentation on request.
07

Function test

Run a test recording, verify firewall log shows zero outbound connections.

Architecture

No data path to the outside

All processing components run exclusively on the workstation. STT (Whisper), speaker diarisation (pyannote), summarisation (Mistral) — local, without API calls, without external servers.

Data privacy

Session lifecycle

Audio data never leaves the workstation. Session data is fully removed after export or deletion. No raw data persistence after completion.

Integration

Compatibility

MDM / Endpoint management

Microsoft Intune · SCCM · baramundi
Silent install

msiexec /quiet — no user interaction
Group policy (GPO)

Registry keys fully documented
Firewall isolation

No outbound connections required
Offline deployment

Air-gap capable. No CDN dependencies.
Export formats

.txt · .md · .docx — generated locally

Documentation

Take the docs with you

Executive Briefing and Technical Briefing as PDF.

Executive Briefing (PDF) Technical Briefing (PDF)

Cloud comparison

Request pilot program

For qualified organizations: technical initial assessment, installation in your environment, guided integration process.