Privacy Policy

1. Information We Process

The bot processes several categories of data during normal operation. Each is described below along with how long it persists.

• Voice audio streams. When the bot joins a voice channel, it receives decrypted voice packets from other participants via Discord's voice gateway (DAVE/MLS end-to-end encryption is enforced between participants; the bot is a participant and holds the session key the same way other members do). Audio is processed in memory by a per-speaker Voice Activity Detector (VAD) and routed to our speech-to-text provider. Raw audio is not written to disk by default and is discarded once each utterance is transcribed.
• Transcripts. Each completed utterance produces a transcript line in the form `[user] [language] text`. Transcripts are posted to the Discord text channel configured by the server operator and appended to an on-disk session file if the server operator has enabled the optional record mode. The server operator controls retention of the Discord channel and the session file.
• Language preferences. Each user's per-bot language priority (set via `/lang`) is stored in a local SQLite database on the bot host. It contains the Discord user ID and the chosen language priority; no other profile data.
• Operational logs. Latency metrics, provider error codes, and VAD queue depths are logged for diagnostics. Logs do not contain transcript text.
• Landing page visits. The kkachi.live landing site is a static site. It does not set cookies, run analytics trackers, or collect IP addresses beyond what the hosting CDN (Cloudflare Pages) retains for its own edge logs.

2. How We Use the Data

We use voice audio and transcripts solely to provide the transcription and translation feature in the Discord channel where the bot has been invited. Language preferences drive automatic detection fallbacks. Operational logs are used to monitor latency, detect provider outages, and debug failures.

• We do not sell, rent, or share personal data with third parties beyond the sub-processors listed below.
• We do not use voice or transcript data to train our own models.
• We do not serve advertising.

3. Sub-processors

To deliver transcription and translation, audio is sent to speech-to-text providers and translation models operated by the following companies. Each sub-processor receives only the minimum data necessary and is bound by their own privacy terms.

• Groq, Inc. — speech-to-text (Whisper-Large-V3-Turbo). Receives short utterance audio segments.
• OpenAI, L.L.C. — speech-to-text and translation fallback (gpt-4o-transcribe, gpt-4o-mini). Receives utterance audio and/or transcribed text for translation.
• Discord, Inc. — the platform hosting the voice channel. The bot operates within Discord's Developer Terms of Service and DAVE E2EE protocol.
• Cloudflare, Inc. — hosts the landing site as a static asset CDN.

4. Retention

• Voice audio: in memory only, discarded within seconds of utterance completion.
• Transcripts posted to Discord text channels: retained for as long as the server operator retains the channel or manually deletes messages. This is outside our control.
• Optional on-disk session WAV + transcript files: retained on the bot host and under the server operator's control.
• Language preferences: retained until the user runs `/lang clear` or the database row is removed.
• Operational logs: retained for up to 30 days for debugging purposes.

5. Security

Voice packets travel over Discord's DAVE end-to-end encrypted voice gateway. The bot obtains session keys as a voice-channel participant, matching the security model of any human participant. Transport to sub-processors uses TLS. Sub-processor API credentials are stored as environment variables on the bot host and are not accessible from the landing site.

6. Your Choices

• Leave the voice channel to immediately stop any processing of your voice.
• Ask the server operator to remove the bot — once removed, the bot retains no residual state for that server.
• Run `/lang clear` to delete your stored language preference.
• Request deletion of operational logs tied to your user ID via the contact channel below.

7. International Transfers

Sub-processors may operate servers in the United States or other regions. By using the service on a Discord server, you acknowledge that transcripts and short voice segments may be processed outside your country of residence under the sub-processor's own data protection terms.

8. Children

kkachi is not directed at children under 13. It inherits Discord's age requirement (13+ in most regions, higher where required by law). We do not knowingly collect data from children under Discord's minimum age.

9. Changes to This Policy

We may update this policy. Material changes will be announced in the kkachi support Discord server and noted in the "Last updated" date above. Continued use of the service after an update constitutes acceptance of the revised policy.