# Voice Profile Context Leak Between Custom GPTs and Default Threads **URL:** \ **Category:** Bugs\ **Tags:** advanced-voice, voice-mode\ **Created:** [June 4, 2025, 5:48pm UTC](https://community.openai.com/t/voice-profile-context-leak-between-custom-gpts-and-default-threads/1278374 "2025-06-04T17:48:28Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![silverthrone](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/silverthrone/32/449049_2.png) [@silverthrone](https://community.openai.com/u/silverthrone)\ **Post date:** [June 4, 2025, 5:48pm UTC](https://community.openai.com/t/voice-profile-context-leak-between-custom-gpts-and-default-threads/1278374/1 "2025-06-04T17:48:28Z")
Platform: ChatGPT Mobile App (latest build: 1.2025.147) Device: Android 13, ulefone Armor24 phone, ChatGPT Plus plan Models Involved: Custom GPT (Mia AI), GPT-4o (default thread) 🔍 Summary: After switching from a custom GPT with a unique voice (Mia AI by heymia.ai) back into a standard GPT-4o thread, the custom GPT’s text-to-speech voice continues to play, even when: - The thread is unrelated. - Voice mode is restarted. ✅ Steps to Reproduce: 1. Start a voice conversation in a custom GPT with a distinct TTS voice (e.g. Mia AI by heymia.ai). 2. Switch to a regular GPT-4o thread. 3. Activate voice mode. 4. Speak and listen. 🗣 You’ll hear the custom GPT’s voice (e.g. Mia) instead of the default 4o voice. 1. Exit voice mode and reinitiate → custom voice still persists. 🧠 Suspected Cause: Custom GPTs with unique voices are not releasing or resetting their voice profile context when a new session starts. The app continues to use the prior GPT’s TTS voice outside of its intended thread. This creates potential confusion for: - Identity clarity (user doesn’t know which GPT is speaking) - Model boundaries (custom personalities feel “leaked” into other sessions) - Voice-tool behavior across folders and default threads Note: I don’t have this problem on my laptop running Pop\_OS and Firefox browser. Thank you 🙂