Voice: A complete solution for audio-based toxicity
GGWP delivers customizable sanctioning rules to automatically and immediately protect your users from toxicity with minimal manual effort
We understand moderation isn’t always black and white. GGWP uses contextualized judgment that considers user history, reports, and conversation details to identify and act on incidents
Efficiently hone in on toxicity by leveraging our industry-leading models. We’ll work with you to find a solution that fits your budget
Our efficient onboarding process and straightforward APIs will have you up and running in around a week, and our team is here to support you at every step.
GGWP filters voice communications through its established text-based models for more efficient and accurate moderation. We synthesize relevant information from user history, voice and chat logs, and user reports to give you full context on each interaction. And with these interactions transcribed, they’ll be more secure and easier to review.
Whether your biggest audience is localized to one continent or scattered across the globe, GGWP can help you keep it safe and engaged. We apply our full suite of features and unique algorithms across a growing list of supported languages: English, French, Italian, German, Spanish, Portuguese, Chinese, Japanese, Korean, Russian, Turkish, and Indonesian.
GGWP was designed to work well right out of the box, but we understand that your needs are as unique as your product. You can customize your GGWP experience with various sensitivity and monitoring settings to properly match the tone of your community. GGWP has its own pre-set blocklists for toxic phrases, or you can customize your own!
We recommend OGG / Opus as it is 60-80% the size of MP3 and 10% the size of WAV files for same duration and quality.
We use probabilistic models and set different confidence thresholds depending on the detection type and application. For more critical and severe detections precision values range from 95-98%. For milder detections in more sensitive domains (eg: games for kids) we optimize for recall, with general values over 95%.
We are also constantly improving our models with a comprehensive QA process that includes manual labelling by our in-house team, incorporating customer feedback on individual incidents, tracking new terms that are starting to appear with more frequency, and adding language that is emerging in the gaming community.
Our scores take into account different sources of information and take a really long term view. They also take into account the confidence and severity of each incident to ensure that reputation impacts are fair and feel reliable for decision making to moderators and can be used by our AutoMod sanctioning system.
GGWP is committed to protecting the privacy and personal data of our customers. We understand the importance of complying with the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA) and have taken necessary steps to ensure that we are fully compliant with both regulations.
As a data processor, we respect and work with data controllers to ensure that their users' data is protected and processed in accordance with GDPR and CCPA regulations. We have implemented appropriate technical and organizational measures to safeguard personal data and comply with the data controller over a given user's right to be forgotten as well as the right for their customer to access their data.
We take our responsibility as a data processor seriously and have implemented strict policies and procedures to ensure that all personal data is processed lawfully, fairly, and transparently. Our team undergoes regular training to ensure that they are fully aware of their responsibilities under GDPR and CCPA.
We accommodate that by default - our models are available by API across regions and our live filtering detections run in <20 ms.