Optional module · also works on its ownTrainexus Video Analysis

Data-driven feedback on presence, voice & speech.

Upload any video and within minutes you receive a detailed assessment: body language, gaze, voice quality and fluency, structured and comparable. It also works alongside a conversation in the Dialogue Simulation.

Go to Video Analysis ↗ Request trial access

Annotated video: skeleton overlay, hand trails and gaze direction.

What is analysed

Four dimensions, one rhetoric score.

Body

Body language & gesture

Skeleton tracking, gesture activity, upper-body stability, arm posture.

Gaze

Head & eye contact

Head orientation, gaze direction, camera contact in per cent, turning towards the presentation.

Voice

Pitch, modulation, volume, voice clarity (HNR), jitter and shimmer.

Speech

Transcript with timestamps, speaking rate (WPM), pauses, filler words.

No emotion or facial-expression recognition: Video Analysis does not infer emotions, moods or inner states (such as nervousness or insecurity). Article 5(1)(f) of the EU AI Act prohibits this in education and the workplace; the corresponding features have been removed.

From the dashboard

Every metric over time; click to enlarge.

Stimm-Analyse: Tonhöhe und Dynamik — Voice: pitch & dynamics

Gestik-Aktivität und -Intensität — Gesture activity

Sprechtempo in Wörtern pro Minute — Speaking rate (WPM)

Head rotation and tilt over time — Head orientation over time

Rhetorik-Score aus acht Kern-Dimensionen — Rhetoric score (0–100, 8 dimensions)

The differentiator: data protection

Processed in Germany; no video leaves the EU.

Uploads run through our Hetzner server in the EU (Finland). The compute-intensive analysis takes place on a dedicated GPU server in Germany (IONOS, Frankfurt am Main data centre). All analysis models run locally on that server; video and images stay in the EU, and nobody but you sees the recording. Details in the privacy policy →

✓Local analysis models on a German GPU server; raw video deleted immediately after processing

✓Optional AI written feedback only on request; only the transcript and metrics go to Claude (zero data retention), never video or images

✓Biometric-adjacent analysis (Art. 9 GDPR) only with consent; a consent gate for other people appearing in the video

In three steps

Upload a video

Any presentation or conversation video, including parallel recordings from the Dialogue Simulation.

Automatic analysis in Germany

Ten analysis modules run in parallel using local models on the German GPU server; the raw video is deleted immediately afterwards.

Structured feedback

Within minutes: a dashboard with annotated video, charts, rhetoric score, benchmark and optional AI coaching, delivered by email link.

ProcessingGPU server in Germany (IONOS, Frankfurt a. M.)

Data pathUpload via Hetzner (EU, Finland) · analysis in Germany · no video or image leaves the EU

AI coachingClaude (zero data retention): transcript and metrics only

DeletionRaw video immediately · results after 14 days or at the click of a button

Third-party footageConsent gate: deleted after 3 days without consent

Available now, data-protection compliant, in Germany.

Upload your video and receive the full evaluation within minutes. Processing runs on local models on a server in Germany; no video and no image leaves the EU. Get in touch.

Go to Video Analysis ↗ Request trial access