@qvac/bci-whispercpp - npm Package Compare versions

Comparing version

+1

-1

bci.js

		@@ -17,3 +17,3 @@ 'use strict'
		// Neural data is ~1 MB/s at 512ch * 50 Hz * 4 B, so 500 MB ~= 8 minutes of
		// signal. The bound matches qvac-lib-infer-whispercpp and protects against
		// signal. The bound matches transcription-whispercpp and protects against
		// runaway producers.
		@@ -20,0 +20,0 @@ const MAX_BUFFERED_BYTES = 500 * 1024 * 1024

+7

-1

CHANGELOG.md

		@@ -8,2 +8,8 @@ # Changelog

		## [0.1.1] - 2026-06-02

		### Changed

		- Bumped the `qvac-lib-inference-addon-cpp` vcpkg dependency to `1.2.1`.

		## [0.1.0]
		@@ -24,3 +30,3 @@
		- C++ native addon (`NeuralProcessor`, `BCIModel`, `BCIConfig`) using the
		`qvac-lib-inference-addon-cpp` framework, with BCI-specific preprocessing
		`inference-addon-cpp` framework, with BCI-specific preprocessing
		(Gaussian smoothing, low-rank day projection, softsign non-linearity) and
		@@ -27,0 +33,0 @@ mel-layout injection into a patched whisper.cpp encoder.

+1

-1

NOTICE

		@@ -71,3 +71,3 @@ @qvac/bci-whispercpp

		qvac-lib-inference-addon-cpp
		inference-addon-cpp
		https://github.com/tetherto/qvac
		@@ -74,0 +74,0 @@ Copyright (c) 2024-2026 Tether Data, S.A. de C.V.

+2

-6

package.json

		{
		"name": "@qvac/bci-whispercpp",
		"version": "0.1.0",
		"version": "0.1.1",
		"description": "Brain-Computer Interface (BCI) neural signal transcription addon for qvac, powered by whisper.cpp",
		@@ -67,6 +67,3 @@ "addon": true,
		"cmake-vcpkg": "^1.1.0",
		"fs": "npm:bare-fs",
		"os": "npm:bare-os@^3.6.2",
		"standard": "^17.1.2",
		"tty": "npm:bare-node-tty",
		"typescript": "^5.9.2"
		@@ -79,4 +76,3 @@ },
		"bare-fs": "^4.5.1",
		"bare-path": "^3.0.0",
		"path": "npm:bare-path"
		"bare-path": "^3.0.0"
		},
		@@ -83,0 +79,0 @@ "exports": {

+2

-2

README.md

		@@ -321,3 +321,3 @@ # @qvac/bci-whispercpp
		- [`whisper_full_params` in whisper.cpp](https://github.com/ggerganov/whisper.cpp/blob/master/include/whisper.h)
		- Concrete shapes used in production: see the [examples](examples) directory and [`@qvac/transcription-whispercpp`](https://github.com/tetherto/qvac/tree/main/packages/qvac-lib-infer-whispercpp) for richer usage patterns (VAD, chunking, live streaming).
		- Concrete shapes used in production: see the [examples](examples) directory and [`@qvac/transcription-whispercpp`](https://github.com/tetherto/qvac/tree/main/packages/transcription-whispercpp) for richer usage patterns (VAD, chunking, live streaming).

		@@ -364,3 +364,3 @@ ## whisper.cpp Patches
		- whisper.cpp fork (Tether): [`tetherto/qvac-ext-lib-whisper.cpp`](https://github.com/tetherto/qvac-ext-lib-whisper.cpp)
		- Sibling package — audio transcription: [`@qvac/transcription-whispercpp`](https://github.com/tetherto/qvac/tree/main/packages/qvac-lib-infer-whispercpp)
		- Sibling package — audio transcription: [`@qvac/transcription-whispercpp`](https://github.com/tetherto/qvac/tree/main/packages/transcription-whispercpp)
		- vcpkg registry: [`qvac-registry-vcpkg`](https://github.com/tetherto/qvac-registry-vcpkg)
		@@ -367,0 +367,0 @@ - BrainWhisperer reference (Python): the model checkpoints converted by `scripts/convert-model.py`

prebuilds/android-arm64/qvac__bci-whispercpp.bare