Copyright © 2020-2026 Traco Systems a.s.

version 2025.10.0

Cloud UX | Traco AI | User Guide

This guide provides instructions for using the Traco AI application to leverage artificial intelligence for media analysis, transcription, translation, and voiceover creation in Avid MediaCentral | Cloud UX.

Table of Contents

Preface

This is an essential guide for the Traco AI application in MediaCentral | Cloud UX, providing information about how to use AI-powered features for media analysis, transcription, translation, and voiceover creation.

Traco AI at your organization may be configured differently than shown in this document. Available features and options will vary according to your system configuration and user permissions. Ask your system administrator if you think something is missing or configuration changes are required.

Introduction

Traco AI is an application created for the Avid MediaCentral platform with the aim of integrating artificial intelligence capabilities for media content analysis and enhancement. Traco AI is available as a standalone application inside Cloud UX and also as a tab in the Asset Inspector for asset-specific AI operations.

The main purpose of Traco AI is to allow users to:

  • Search media content using natural language queries with semantic understanding
  • Create voiceovers from text using AI-powered text-to-speech technology
  • Transcribe audio content with speaker identification and timing information
  • Translate transcriptions into multiple languages with live translation support
  • Generate summaries of media content for quick content overview
  • Recognize and track persons appearing in video content
  • Detect content categories for automatic metadata tagging
  • Export subtitles in multiple formats (TXT for Avid MediaComposer, WebVTT, SRT)

How to Access Cloud UX | Traco AI

Every user needs their own login name and password to access Cloud UX. After signing in, each user with Traco AI entitlement can access Traco AI features through two different interfaces, depending on the task at hand.

Asset Inspector | Traco AI Tab

When working with a specific asset (masterclip or subclip) in the Asset Inspector, users can access the Traco AI tab to perform asset-specific AI operations such as transcription, translation, person recognition, and category detection.

Cloud UX | Traco AI Application

Users can open the Traco AI standalone application from the Cloud UX applications menu bar. This application provides access to semantic search, text-to-speech voiceover creation, and global jobs management.

Asset Inspector | Traco AI Tab Interface

The Asset Inspector | Traco AI Tab provides AI-powered analysis features that are specific to the currently opened asset. This interface appears as a tab in the Cloud UX Asset Inspector when viewing a supported asset (masterclip or subclip from Interplay PAM).

The Traco AI tab provides four main feature categories accessible through the tab header:

  • Speech (Transcription, Translation, Summary)
  • Persons (Face Recognition, Markers)
  • Categories (Content Classification)
  • Jobs (Asset-specific job monitoring)

Integration with Cloud UX Search: AI-generated data such as detected persons, categories, and summaries are automatically stored as custom Avid PAM attributes. This makes them available in the Cloud UX Search application for searching and filtering. Traco AI includes an automatic update mechanism that synchronizes any changes made to AI data (whether by automatic processes or manual user edits) to the corresponding PAM attributes, ensuring that Cloud UX indexed search always has up-to-date values.

Working with Speech

The Speech section provides comprehensive tools for transcribing, translating, and summarizing audio content from your media assets.

Starting a Transcription

To transcribe the audio content of an asset:

  1. Open a masterclip or subclip in the Asset Inspector
  2. Click on the Traco AI tab
  3. Ensure the Speech page is active (click "Speech" in the header if needed)
  4. Click the + button to open the transcription options
  5. Configure transcription settings:
    • Language: Select the language spoken in the audio
    • Audio Channel: Select which audio channel(s) to transcribe
    • Engine: Select the transcription engine to use
    • Auto-translate: Optionally select a language for automatic translation
  6. Click "Start Transcription"

The transcription job will begin processing. You can monitor its progress in the Jobs section.

Viewing and Editing Transcriptions

Once a transcription is complete:

  1. Select the transcription from the speech dropdown
  2. The transcription appears with:
    • Speaker labels - Identification of different speakers
    • Word-level timing - Each word is synchronized with the audio
    • Synchronized highlighting - Words highlight as the video plays

To edit transcription text:

  1. Click the "Edit" button
  2. Click on individual words to modify them
  3. Use Save to confirm changes or Cancel to discard
  4. Edited transcriptions maintain their timing information

Note: Pressing Enter while editing a word will save the change without creating a new line.

Text Highlighting and Playback Synchronization

The transcription text automatically highlights in sync with video playback:

  • Words are highlighted as they are spoken in the video
  • Click on any word to jump the video playback to that time
  • Enable Auto Scroll to automatically scroll the text as the video plays
  • Use the Highlight button to toggle synchronized highlighting on/off

This feature makes it easy to verify transcription accuracy and navigate through your content.

Searching Within Transcriptions

To find specific words or phrases in a transcription:

  1. Click the search icon
  2. Enter your search term
  3. Matching words are highlighted throughout the transcription
  4. Click on highlighted results to jump to that point in the video

Working with Translations

The Translation feature allows you to translate transcribed content into multiple languages with real-time updates as transcription progresses.

Translating Transcriptions

To translate a transcription:

  1. Ensure you have an existing transcription
  2. Click the "Translate To" dropdown
  3. Select the target language
  4. Click "Translate"

The translation job will begin processing. Once complete, the translated text appears with the same timing and speaker information as the original transcription.

Live Translation

If you select a translation language when starting a new transcription (using the auto-translate option), the translation will be created in real-time as the transcription progresses. This is called "live translation."

Live translation allows you to:

  • See translated text appear as transcription progresses
  • Work with translations without waiting for full transcription completion
  • Maintain synchronized timing between original and translated text

Viewing and Editing Translations

To view and work with translations:

  1. Select the desired translation from the translations dropdown
  2. The translation appears with:
    • Speaker labels
    • Synchronized highlighting during playback
    • Word-level timing

To edit translation text:

  1. Click the "Edit" button
  2. Make changes to individual words or segments
  3. Save to confirm or Cancel to discard changes

Use the "Back to Transcription" button to return to viewing the original transcription.

Managing Multiple Translations

You can create multiple translations of the same transcription:

  1. Select different languages from the "Translate To" dropdown
  2. Each translation is saved separately
  3. Switch between translations using the translations dropdown
  4. View different language versions while maintaining the same timing

When creating a summary, the system intelligently links summaries with translations—changing the summary display language will automatically switch to the corresponding translation if available.

Working with Summaries

AI-generated summaries provide quick content overviews of your transcribed media. Summaries are stored as custom Avid PAM attributes, making them searchable in the Cloud UX Search application.

Creating Summaries

To create a summary:

  1. Ensure you have an existing transcription
  2. Click "Create Summary"
  3. The system analyzes the transcription and generates a concise summary

Summaries are:

  • Created from the transcribed content
  • Available in multiple languages
  • Automatically linked to translations

Translating Summaries

To translate a summary:

  1. Ensure you have an existing summary
  2. Click "Translate Summary"
  3. Select the target language
  4. The summary is translated and linked to the corresponding transcription translation

When you change the summary display language, the system automatically switches to the corresponding transcription translation, ensuring consistency across all content.

Exporting Subtitles

You can export transcriptions and translations as subtitle files in multiple formats for use in video editing workflows.

To export subtitles:

  1. View the transcription or translation you want to export
  2. Click the "Download Subtitle" button
  3. Select subtitle format (if multiple formats are available)
  4. Choose options:
    • Include speaker labels
    • Include timecode
  5. The subtitle file is downloaded to your computer

Subtitle files include:

  • Timing information synchronized with the media
  • Speaker identification (if enabled)
  • Text content in the selected language

Supported subtitle formats include TXT for Avid MediaComposer, WebVTT, and SRT. These files can be imported into Avid editing applications and other video editing software for subtitle creation.

Person Recognition and Markers

The Person Recognition feature uses AI to detect and identify persons appearing in your video content. Detected persons are stored as custom Avid PAM attributes, making them searchable in the Cloud UX Search application.

Starting Face Recognition

To start face recognition on an asset:

  1. Open a masterclip or subclip in the Asset Inspector
  2. Click on the Traco AI tab
  3. Click "Persons" in the header
  4. Click "Start Face Recognition"

The face recognition job will analyze the video and identify persons throughout the content.

Creating Timeline Markers

Once face recognition is complete, you can create timeline markers for detected persons:

  1. View the list of detected persons
  2. Each person entry shows:
    • Thumbnail of the detected face
    • Time ranges where the person appears
    • Detection confidence
  3. Click "Create Markers" for a specific person or all persons
  4. Markers are created in the Asset Inspector timeline

Markers:

  • Use different colors for different persons
  • Appear at the time points where each person was detected
  • Help quickly navigate to sections featuring specific individuals
  • Are visible in the Asset Editor timeline

Category Detection

The Category Detection feature automatically analyzes video content and assigns relevant category tags. Detected categories are stored as custom Avid PAM attributes, making them searchable in the Cloud UX Search application.

Starting Category Detection

To start category detection:

  1. Open a masterclip or subclip in the Asset Inspector
  2. Click on the Traco AI tab
  3. Click "Categories" in the header
  4. Click "Start Category Detection"

The category detection job will analyze the video content and identify relevant categories.

Viewing Detected Categories

Once category detection is complete:

  1. View the list of detected categories
  2. Each category shows:
    • Category name
    • Confidence score
    • Relevance to the content

Categories help with:

  • Media organization
  • Content discovery
  • Metadata enrichment
  • Automated classification workflows

Detected categories can be used for creating metadata, organizing assets, or building automated workflows based on content type.

Asset-Specific Jobs

The Jobs section in the Traco AI tab shows jobs that are specific to the currently opened asset.

To view asset jobs:

  1. Open an asset in the Asset Inspector
  2. Click on the Traco AI tab
  3. Click "Jobs" in the header

The jobs list displays:

  • Transcription jobs for this asset
  • Translation jobs for this asset
  • Face recognition jobs
  • Category detection jobs
  • Job status and progress information
  • Error details for failed jobs

This view helps you monitor AI processing progress for the specific asset you're working with, separate from the global jobs view in the standalone Traco AI application.

Cloud UX | Traco AI Application Interface

The Cloud UX | Traco AI application is a standalone application that runs independently of any specific asset. It provides three main features accessible through the application header:

  • Search
  • Text to Speech (if enabled by administrator)
  • Jobs

The Semantic Search feature allows you to find media assets using natural language queries, making it easy to discover content based on meaning rather than just keywords.

To search for media assets:

  1. Open the Traco AI application from the Cloud UX applications menu
  2. Ensure the Search page is active (click "Search" in the header if needed)
  3. Enter your search query in the search box using natural language
  4. Click the search button or press Enter

The system will analyze your query and return relevant assets ranked by relevance.

Using Search Filters

You can refine your search results using date filters:

  1. Click the date filter icon
  2. Select a date range:
    • From date: Select the earliest date to include in results
    • To date: Select the latest date to include in results
  3. Click Apply to update the search results

Viewing Search Results

Search results are displayed with:

  • Thumbnail preview - Visual representation of the asset
  • Relevance highlighting - Results are highlighted based on relevance score
  • Asset information - Name, duration, and other metadata
  • Relevance score - Visual indicator showing how well the asset matches your query

Results are automatically ranked by relevance, with the most relevant assets appearing first.

Working with Search Results

You can interact with search results in several ways:

  • Click on a result to highlight and select it
  • Drag and drop results to create new sequences or add to timeline
  • View details by clicking on the asset name
  • Filter by subclips - Search results include both master clips and subclips

When dragging search results to create sequences, the system will automatically handle timing information and asset references.

Text-to-Speech and Voiceover Creation

The Text-to-Speech feature allows you to create high-quality voiceovers from text using AI-powered voice synthesis. This feature is only available if enabled by your administrator.

Creating a Voiceover

To create a voiceover from text:

  1. Open the Traco AI application
  2. Click "Text to Speech" in the header
  3. Enter or paste your text in the text input area
  4. Configure voiceover settings:
    • Select the voice/speaker
    • Select the language
    • Optionally customize the GPT prompt for text refinement
  5. Click "Create Voiceover" to generate the audio

The system will process your text and create a voiceover asset that can be used in your productions.

Customizing Voice and Language

The text-to-speech system supports multiple voices and languages:

  • Voice selection: Choose from available voice profiles
  • Language selection: Select the target language for the voiceover
  • Prompt customization: Use custom GPT prompts to refine the text before conversion

Creating Sequences with Voiceovers

You can create complete sequences from search results with automatically generated voiceovers:

  1. Perform a search to find relevant media assets
  2. Select one or more search results
  3. Drag the selected results to create a new sequence
  4. The system can automatically generate voiceovers from summaries of the selected assets

This feature is particularly useful for quickly creating narrated sequences from existing media content.

Jobs Management

The Jobs page provides a global view of all AI processing jobs submitted by you across the system.

Viewing Jobs

To view your jobs:

  1. Open the Traco AI application
  2. Click "Jobs" in the header
  3. The jobs list displays all your AI processing jobs

Each job entry shows:

  • Job Name - Descriptive name of the job
  • Job Type - Type of AI operation (Transcription, Translation, Face Recognition, etc.)
  • Status - Current state of the job (Pending, Running, Completed, Failed)
  • Progress - Visual indicator of job completion percentage
  • Time - Submission and completion timestamps
  • Asset - Link to the asset being processed

Monitoring Job Status

Jobs can be in various states:

  • Pending - Job is queued and waiting to start
  • Running - Job is currently being processed
  • Completed - Job finished successfully
  • Failed - Job encountered an error

Click on a job to view detailed error information if the job has failed.

Filtering and Searching Jobs

You can manage your jobs list:

  • Pagination - Navigate through multiple pages of jobs
  • Page size - Adjust how many jobs are displayed per page
  • Status filtering - Filter jobs by their current status
  • Search - Search for specific jobs by name or type