This repository packages the current Kokoro 82M LiteRT/TFLite runtime used by the Reachy edge robot-agent project.

It is sourced from hexgrad/Kokoro-82M and contains the accepted text-to-decoder-input frontend bucket plus the accepted merged decoder/vocoder graph.

The runtime still uses the kokoro Python package for KPipeline.g2p() and KPipeline.en_tokenize(). It must not instantiate Kokoro KModel in the request path. Neural inference is served by the LiteRT frontend bucket and the LiteRT decoder/vocoder.

The current frontend bucket is T=48, with max 128 decoder frames and 256 F0/noise frames. Longer or multi-segment text must be deterministically chunked and repacked before inference.

The package includes custom op builds for local Linux x86-64 development and Jetson/Linux aarch64 deployment:

The aarch64 binary was cross-compiled from custom_ops/kokoro_source_stft_custom_op_native.cc with:

The expected aarch64 SHA-256 is recorded in kokoro_litert_manifest.json under decoder_vocoder.custom_op.linux_aarch64_sha256. Jetson target-device loading and synthesis benchmarking are still required.

Frontend bucket acceptance is recorded in:

The local acceptance result for this package:

Decoder/vocoder acceptance is recorded in:

The merged decoder is a one-interpreter graph connected through the KokoroSourceStft custom op. The custom op remains a CPU custom-op island unless implemented as a GPU-capable custom kernel or delegate.

In the Reachy robot-agent repo:

Expected output is a mono 24 kHz WAV file.

The upstream Kokoro model card lists hexgrad/Kokoro-82M under Apache-2.0. This converted runtime package is distributed under Apache-2.0 as a derived runtime form. See LICENSE and NOTICE.