Announcing Unitary Standard, our latest off-the-shelf classifier

We have a few exciting product updates to share today.

🫴 New Product: Unitary Standard

Unitary Standard classifies images, videos and any text like titles and descriptions against the following categories of harmful content:

• Adult & Sexual. Explicit sexual activity, nudity, detailed explanations of sexual encounters…
• Non-medical Drugs. Pills, coke, marihuana, paraphernalia, minors drinking, minors smoking…
• Violence and Injury. People killing or sexually assaulting other people, violent fights in uncontrolled settings, animal abuse, physical threats, suicide, self-injury or eating disorders and their glorification, graphic injuries…
• Weapons and Firearms. Firearms, ammunition, explosives, non-kitchen knives or homemade weapons in dangerous settings, instructional content on weapons…
• Hate Speech and Hate Symbols. Symbols from hate organisations, discrimination of people based on their identity…

Unitary Standard works off-the-shelf. You can start getting value out of it as soon as we give you an API key.

Unitary Standard goes beyond basic Items and Characteristics classification. It uses all modalities, from speech to texts within the images, to determine if a piece of content is harmful.

As a reminder, we are also offering Unitary Premium, which provides further accuracy than Unitary Standard by fine-tuning our models to your platform's data and content guidelines. Please let us know if you’d like to explore an evaluation of Unitary Premium that can give you further classification performance.

📚 A new simplified version of Unitary’s API documentation — filled with examples to help you integrate Unitary's API

We’ve improved our API documentation to make the onboarding to Unitary's API even simpler, including:

• A new Standard & Premium: Policy Classification endpoints page, where you can learn how to integrate the new Standard and Premium products.
• A features and requirements page that answers the most frequently asked questions.
• Even more code examples across the docs to help you integrate with Unitary’s API.

💬 Unitary products can now process speech in many more languages

All Unitary products are now processing speech in many more languages.

Additionally, performance for English has slightly improved.

The API add-on to Include speech audio transcriptions in the API response will incorporate two new fields to help you detect any language spoken in the video.

• language, which will contain the main detected language in ISO 639 format e.g. “EN” for English or “ES” for Spanish
• language_probability, which will contain the confidence in the language being detected e.g. 0.7 for a 70% confidence

These two new fields will be included under a new speech group field that will also contain the field texts, which returns the actual speech, as seen in the example below

{
"speech":{
"texts":["This is speech", "in a video"],
"language":"en",
"language_probability":0.7
}
}


🚄 New add-on: sync endpoint for image classification

We know that speed is important for many of your use cases so we are now offering a synchronous response for our image classification endpoint when it targets sub-second latency. This removes the need to implement webhooks for images.

Let us know if you’d like to know more about any of these updates!