What is the Speaker Index?
Intelligent labelling of voices without AI training.
The Speaker Index of the Deep Indexer allows you to assign a unique identifier to individuals in video and audio material, making it easier to translate them into speaker metadata. The Speaker Index also enables clustering and manual labeling of voices in your media archive, so you can build a dataset using your existing audio-visual assets without the need to upload additional training material.
Key Functionality:
- Voice Indexing:
Once voices are detected and recognized, they are indexed, allowing users to search for specific speakers or filter content based on who is speaking in the media. This creates a searchable database of voices across a wide collection of audio and video files.
The Voice Index provides the easiest way to manage unknown voices. Each voice is automatically assigned a unique ID, allowing you to rename it instantly. Typically, in the Deep Model Customizer, you would need to upload training material for each voice, but with the Voice Index, voices are instantly recognizable without requiring training data.
Accessing the Deep Indexer:
The Deep Indexer is triggered together with the Speaker Identification Module of the Deep Media Analyzer, once the user decides to activate it in the configuration of the parameters.
To access the Voice Index results, the user can use the "Indexes" button - which displays a fingerprint - in the left-hand menu.
How can I trigger the Face Indexing?
- Select the Media File: Choose the media file you want to analyze.
- Activate the Speaker Identification Module: In the left column, select the "Speaker Identification" module.
- Define the Model & Parameters: Choose the model for analysis from the available options, set the parameters, and click the yellow "Add Module" button.
- Triggering the Deep Indexer:
Activate Numbering of Labels for Unknown Identities: Enables the Deep Indexer to assign unique IDs to unidentified faces across all assets. Renaming an indexed face will automatically update previously indexed images and future analyses. - (Optional) Cluster Unknown Identities: Groups unrecognized voices as "unknowns" without assigning IDs. All unrecognized individuals will remain unnamed.
- Start the Analysis: You can either add more modules or begin the analysis immediately by clicking "Start Analysis"
How can I access the Voice Indexes?
To access the Voice Index, the user can use the "Indexes" button - which displays a fingerprint - in the left-hand menu.
In the Indexes Collection section you see all currently available Index Collections, this can be either Face Indexes or Voice Indexes. Per default you have one Index per type, but you are able to create several indexes per type.
Index Collection Card:
Each Index summarizes the most important information in its Collection Card,
this information features:
- Name: Name of the Index Collection
- Type: Face or Voice
- Date Created
- Date Last Modified
- Number of Identities in the Collection
-
Description: User given description of the Collection
By clicking "Open" you access the specific Index Collection.
Result Cards
Results are displayed as cards in chronological order with a small identicon image and the respective Index number.
By clicking on the Three dots (...) in a card you can edit each Indexed Voice:
- Edit: Manually naming / labeling the ID for example with its name.
- Details: Opening the voice sample.
- Delete: Permanently deleting this Index
Note:
Deleting the Indexed Voice does not prevent the Voice to be indexed in the next analysis of the same asset or in the analysis of another asset.
Identicons:
Identicons are simple, unique images created from data like an email or username. They use patterns of colored squares to give each user a distinct picture, often used on websites and apps as profile images when no photo is provided.
Search Field:
Located in the top bar, the search field includes filter settings for refining your results.
- Name field: Enter a name to view results that either match or don't match the entered name.
- Labeled (Yes/No): Only find manually labeled results
After adjusting filters, click "Apply" to apply them. Active filters appear in a black box beneath the search field and can be cleared by clicking the X symbol.