anyformer 0.4.15 → 0.4.16

package/CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## 0.4.16 — 2026-10-30
+
+### Changed
+- diskformer pin 7ffbea5 → 93193d0 (v0.6.14: LFM3-Omni speech out runs at 1.5–1.8x real time, up from 0.54x, on an
+  M4 Pro; the inverse STFT uses an FFT with the same output; changes only `engines/lfm3_omni.js`).
+- Browser test `omni-speech` on the new pin (Chrome, this Mac, local files, about 22:50 IST): 13.52 s of speech in
+  7.1 s (1.9x real time), first sound after 0.98 s; abort at the first speech chunk; text reply after it equal.
+
 ## 0.4.15 — 2026-10-30
 
 ### Fixed

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "anyformer",
-  "version": "0.4.15",
+  "version": "0.4.16",
   "description": "One call to run a model in a browser tab: anyformer reads the Hub repo, probes the browser, and picks the back end (transformers.js, LocalMind's WebGPU kernels, diskformer, kohra) that fits the model's format, architecture and your GPU budget.",
   "type": "module",
   "main": "./index.js",

package/src/sources.js

@@ -9,6 +9,8 @@ export const PINS = {
   // v0.3.1: seeded block-KV graphs (Nemotron-Labs-Diffusion-3B), chatFormat, split external data; download progress
   // (progress_callback) and the graph files cached in Cache Storage 'kohra-models'.
   kohra: 'bf95d44c3e3cfc001201baa03fa121c2bf9c7d73',
+  // v0.6.14 (93193d0): LFM3-Omni speech out's inverse STFT uses an FFT (same samples), 0.54x → 1.5–1.8x real time on
+  // an M4 Pro. Changes only engines/lfm3_omni.js.
   // v0.6.13 (7ffbea5): LFM3-Omni serves one reply at a time; overlapping runs broke onnxruntime-web WebGPU.
   // v0.6.12 (1fe19f8): LFM3-Omni speech out (interleaved text and 24 kHz speech; depth and codec decoders). Changes
   // only engines/lfm3_omni.js; its text, image and speech-in gate (3/3) is unchanged.
@@ -19,7 +21,7 @@ export const PINS = {
   // Gemma 4 E2B/E4B and Ternary Bonsai 2) under engines/, and the llama.cpp path under ggml/, which loads split GGUFs
   // and drafters, and reads past 2 GiB (Nemotron 3.5 Bolt E4B needs v0.6.8 or later). The hand engines' ingest
   // streams the GGUF header with progress and resumes a dropped range.
-  diskformer: '7ffbea54a5f09494e1f6c970cce7c4c7b1b8460d',
+  diskformer: '93193d0913f5e76059fa45db509dc670e7527430',
 };
 
 const gh = (repo, sha, path) => `https://cdn.jsdelivr.net/gh/NakliTechie/${repo}@${sha}/${path}`;