Skip to main content

Data Collection

Overview​

The Sensay Data and Voice Collection System is designed to capture comprehensive information about an individual to create an accurate and lifelike digital replica. Our system employs a multi-modal approach, integrating various data sources to build a holistic representation of a person's knowledge, experiences, and personality.

Data Collection Methods​

1. Text-Based Input​

  • Long-form Text Input: Users can provide written responses to a series of prompts designed to capture their thoughts, opinions, and experiences on various topics.
  • File Upload: Documents, essays, or other written works can be uploaded to provide additional context and information.
  • API Integration: For corporate clients, we offer API integrations to collect data from internal knowledge bases, email systems, and collaborative platforms like Slack or Microsoft Teams.

2. Voice Input​

  • Voice Recordings: Users can record spoken responses to prompts, allowing us to capture speech patterns, intonation, and verbal mannerisms.
  • Existing Audio Content: We can process previously recorded audio content such as podcasts, interviews, or speeches.

3. Visual Input​

  • Photo Upload: Users can upload photos to help with visual representation in video-capable replicas.
  • Video Input: Short video clips can be uploaded to capture gestures, expressions, and overall demeanor.

4. External Source Integration​

  • Social Media: With user permission, we can analyze social media posts to understand public-facing personality and interests.
  • YouTube Content: For creators or public figures, we can process their YouTube videos to gather additional data.
  • Published Works: For authors or academics, we can integrate data from their published works.

Voice Collection Process​

  1. Initial Recording Session: Users participate in a guided recording session, speaking on various topics to provide a baseline for voice replication.

  2. Prompt-based Recordings: Users are given a series of prompts covering different subjects, emotions, and speech styles to capture a wide range of vocal expressions.

  3. Natural Conversation Sampling: We record natural conversations (with consent from all parties) to capture authentic speech patterns and interactions.

  4. Phoneme Coverage: Our system ensures comprehensive coverage of all phonemes in the user's language to enable accurate speech synthesis.

  5. Emotion and Intonation Mapping: We collect samples of different emotional states and intonations to replicate the nuances of human speech.

Data Processing and Security​

  1. Data Encryption: All collected data is encrypted using AES 256-bit encryption for storage and TLS for data in transit.

  2. Anonymization: Personal identifiers are separated from the core data used for replica training to enhance privacy.

  3. Selective Processing: Users can review and selectively approve which data is used in the replica creation process.

  4. Regular Audits: We conduct regular security audits to ensure the integrity and protection of all collected data.

Ethical Considerations​

  • Informed Consent: We obtain explicit consent for all data collection and usage.
  • Data Ownership: Users retain ownership of their data and can request its deletion at any time.
  • Transparency: We provide clear information about how the data will be used in the replica creation process.
  • Ongoing Control: Users can update or modify their data inputs over time to ensure their replica remains accurate and up-to-date.

Continuous Improvement​

Our data and voice collection system is continuously evolving. We regularly update our methodologies based on the latest advancements in AI and machine learning to improve the accuracy and capabilities of our digital replicas.

For any questions or concerns about our data and voice collection process, please contact our support team.