IndexTTS 1.5 community integration proposal: reference-audio-only speech requests
Proposing an IndexTTS 1.5 community example for Rapid-MLX's local speech API on Apple Silicon, using the MLX port through mlx-audio.
Unlike F5's paired reference audio/transcript contract, this adapter requires reference audio and allows the request without a reference transcript. The linked tests cover that distinction, forwarding only the appropriate generation fields, and resolving tokenizer/weights without mutating cached configuration.
A ready-to-run example should demonstrate the reference-audio request on the exact 1.5 pack and retain valid output audio plus dependency versions. It should not imply IndexTTS 2 compatibility or arbitrary voice selection. The proposal is a model-specific API note so that clients send the correct conditioning fields, with synthesis validation still to be completed.
Concrete catalog entry: indextts-1.5 → mlx-community/IndexTTS-1.5@d163f13bc1816c20bd79d730cc88f866a2a43ceb. This is a conversion/distribution artifact; it is not presented as an official upstream release.
Source references (fixed Rapid commit): catalog, IndexTTS request and loader tests, audio API guide.
Verification status: this proposal is based on reading the catalog, implementation and the linked documentation/tests. No fresh model inference or test-suite run was performed for this submission; source tests are not an end-to-end qualification. It requests a community documentation scope, not an official endorsement or a performance listing.
Disclosure: I maintain Rapid-MLX. Prepared with AI assistance under my direction. Anonymous usage-metadata telemetry is enabled by default; DO_NOT_TRACK=1 disables it. The original model and conversion terms continue to apply.