Wildmoka empowers users to integrate with Artificial Intelligence/Machine Learning services to enrich their live and file-based workflows. Wildmoka comes with native AI capabilities for Intelligent Clipping and Auto-Reformatting (Auto ReZone), and we support integration with third-party transcription, translation, object detection, and metadata enrichment services. All use of AI/ML services is optional, and customer administrators of the Wildmoka customer environments can choose to allow access to these features on a per-user level.
Wildmoka comes with some services configured by default, but customers can choose to disable these and in some cases use their own accounts with the supported AI services.
Transcription & translation
Wildmoka offers built-in transcription and translation services that can be activated for any stream or file ingest.
Transcription: Wildmoka’s automatic speech recognition (ASR) runs directly on the Wildmoka platform, ensuring seamless integration with your production workflows.
Translation: For translation, the system leverages OpenAI running on Microsoft Azure. This configuration ensures enterprise-grade security and strict data privacy, as it does not share customer data with public models or use it for training.
Customers who prefer specific ASR engines can seamlessly connect third-party services, including Deepgram, Speechmatics, and Gladia.
Note: For the very low latency requirements of transcription and translation on live production, Wildmoka relies on Speechmatics.
Image Analysis
To enable automatic scene detection, object detection, and tracking, Wildmoka utilizes open-source image analysis models running directly on the Wildmoka platform.
These models run locally within the Wildmoka infrastructure and no data is shared between customers.
This native image processing is the engine behind features such as Auto ReZone. By tracking objects and action in real-time, Wildmoka can automatically crop and reformat landscape video into vertical formats ("verticalization") for mobile platforms without manual editing.
LLM-Automated Metadata
The LLM-Automated Metadata feature is designed to automatically populate metadata fields based on a prompt and contextual content. This can be activated independently for any particular publication field of any destination and it can be configured with a prompt.
Training
Wildmoka, and the services we use (including OpenAI on Azure), do not use any customer content to train any AI models.