Feature request: connection-independent Read Aloud for already-delivered ChatGPT responses
Nobody has claimed this yet.
- Dominant language
- Rust
- Stars
- 125k
- Forks
- 19.4k
- PR merge metrics
- PR metrics pending
Description
Summary
ChatGPT Read Aloud should not require continuous Internet connectivity once the response text being read has already been delivered to and displayed on the user's device.
This is a platform-wide ChatGPT request, not an iPhone-specific request. It should apply where technically feasible across iOS, Android, Windows, macOS, and other ChatGPT clients.
Problem
When a long completed response is already present on the device, Read Aloud can still degrade or stop as network quality weakens. If connectivity drops completely, playback may stop and fail to resume automatically when the network returns, forcing the user to restart playback.
This creates an avoidable failure mode for users traveling through intermittent connectivity: cars, buses, trains, aircraft with intermittent Wi-Fi, ferries, rural areas, weak cellular zones, unreliable Wi-Fi, and remote-work scenarios.
Interactive Voice understandably requires an active connection to OpenAI. Read Aloud of already-generated static text is a different case: the text already exists locally on the client.
Requested behavior
- Allow locally available response text to be synthesized on-device using a downloadable/local ChatGPT voice or offline speech model where practical.
- Alternatively, allow completed responses to have Read Aloud audio pre-rendered, buffered, or cached so playback can continue through temporary connectivity loss.
- Preserve playback position across network interruptions.
- Automatically resume playback from the same position if connectivity is temporarily required and later returns.
- Do not require a user to manually restart Read Aloud after a transient network loss.
- Implement consistently across supported mobile and desktop clients where technically feasible.
- Consider support for reading completed responses from paired remote/local Work or Codex sessions: a paired mobile client could receive the completed response text and synthesize it locally without requiring the entire local conversation to become a cloud chat.
Reproducible example
A long story response is fully generated and visible on the device. Read Aloud begins normally with good connectivity. As connectivity weakens, playback breaks up; when connectivity is lost it stops; when connectivity returns, playback does not recover automatically and the user must restart Read Aloud.
Why this matters
The key requirement is not a driving-specific feature. It is resilience for any user moving through areas where Internet access is intermittent or unavailable. Once the text has already been delivered to the device, continuous network dependence for basic text-to-speech should be minimized or eliminated.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue names no Codex files, tests, or entry points; first confirm whether this repository owns Read Aloud or whether the work belongs in separate ChatGPT clients. Scope the supported clients and current playback implementation before defining completion around offline or buffered playback, preserved position, and recovery after connectivity loss.
Written by the indexing model from the issue text.
Assessment
- Domain
- audio-video-rtc, desktop-dev, mobile-dev, networking
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100