Privacy Policy

Last Updated: January 27, 2026

Overview

VeilSpeak is a private voice-based AI practice app designed to help you build confidence and prepare for interviews. We prioritize your privacy above all else.

Audio Recording and Processing

What We Collect

VeilSpeak uses the android.permission.RECORD_AUDIO permission to:

• Record your voice during practice sessions
• Enable real-time voice conversations with AI

How We Protect Your Privacy

1. On-Device Voice Modification: Your microphone audio is immediately modified using WSOLA pitch shifting (3 semitones up) with subtle noise injection to anonymize your voice characteristics
2. Local Processing Only: This voice modification occurs entirely on your Android device
3. Anonymized Transmission: Only the modified, anonymized audio is sent to our backend services
4. No Storage: We do not store audio recordings or transcripts on our servers
5. Streaming Only: Audio is processed in real-time and discarded after processing

Data We Collect

Account Information

• Email address (when you sign in with Google)
• User ID (for authentication purposes)

Usage Information

• Session metadata (start/end times)
• Cost tracking for AI service usage
• Error logs (without audio content)

What We Do NOT Collect

• Raw microphone audio
• Your original, unmodified voice
• Audio recordings or transcripts
• Personal conversations or practice content
• Location data
• Contact lists
• Device identifiers beyond what's necessary for authentication

Third-Party Services

VeilSpeak uses the following third-party services:

• Google Sign-In: For authentication (subject to Google's Privacy Policy)
• Firebase Authentication: For secure user authentication
• Speech-to-Text Services: Only anonymized audio is sent
• AI Language Models: Only text transcripts (from anonymized audio) are processed
• Text-to-Speech Services: For generating AI responses

Data Security

• All network communication uses secure WebSocket (WSS) and HTTPS protocols
• Authentication tokens are stored securely using Android's EncryptedSharedPreferences
• Backend services are stateless and do not retain conversation data

Your Rights

You have the right to:

• Delete your account at any time
• Request information about data we've collected
• Stop using the app and have your account data removed

Children's Privacy

VeilSpeak is not intended for children under 13 years of age. We do not knowingly collect data from children.

Changes to This Policy

We may update this privacy policy from time to time. We will notify you of any changes by posting the new policy with an updated "Last Updated" date.

Contact Us

Technical Details

For transparency, here's how audio processing works:

1. Capture: 48 kHz mono PCM-16 audio (20ms frames)
2. Modify: WSOLA pitch shifting + noise injection (on-device)
3. Downsample: 16 kHz for efficient transmission (on-device)
4. Transmit: Only anonymized, downsampled audio via encrypted WebSocket
5. Process: Backend performs speech-to-text on anonymized audio
6. Discard: Audio is immediately discarded after transcription