Repository navigation
voice: stream Paradee model chunks with bounded delivery - #82
Merged
Merged
Conversation
Assisted-by: Codex
Assisted-by: Codex
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Paradee currently accumulates every model chunk before returning audio. Add an ABI 19 stream entry that delivers each model chunk before starting the next, in synchronous windows of at most 24,000 samples. Callback backpressure and native cancellation stop further delivery; existing batch calls remain unchanged. This is model-chunk streaming, not incremental decoding within one model forward pass.
The final callback occurs only after successful synthesis; the return code remains authoritative. Callers must flush queued playback after cancellation/error. No consumer pin, model, admission flag or device-acceptance claim changes here.
Validation: standalone ordered-delivery/bounds/cancellation tests pass, including release assertions; the full FFI contract test translation unit passes host syntax compilation. Added FFI-double coverage for byte-equivalent stream/batch PCM, cancellation, empty output and invalid callbacks, with a focused hosted build/test workflow. Hosted focused contract run 38678390499 compiled/linked the current FFI and passed both suites, including early delivery before the second model inference, native and callback cancellation, and final-callback rejection. General Windows configure encountered a BoringSSL upstream HTTP 503 before source compilation; other broad workflows remain pending. No engine/model rebuild or physical audio test locally.
AI-assisted implementation under the owner's explicit repository completion request. Supports AlphaCompute/alphaphone#423; JNI/native playback integration follows separately after review.