Licenses
Weeve Tools is built on open models, libraries, and fonts. They are listed here with their licenses.
NVIDIA Parakeet TDT 0.6B v3
CC-BY-4.0The speech recognition model that turns your audio into text.
huggingface.co/nvidia/parakeet-tdt-0.6b-v3parakeet.js
MITRuns the Parakeet model in the browser on WebGPU and WebAssembly.
github.com/ysdede/parakeet.jsONNX Runtime Web
MITThe inference engine that executes the models on your device.
onnxruntime.aisherpa-onnx
Apache-2.0The speaker diarization engine that separates who spoke when.
github.com/k2-fsa/sherpa-onnxpyannote segmentation 3.0
MITDetects speech segments and speaker changes in the recording.
huggingface.co/pyannote/segmentation-3.03D-Speaker CAM++ embeddings
Apache-2.0Creates the voice fingerprints used to tell speakers apart.
github.com/modelscope/3D-SpeakerMediabunny
MPL-2.0Packages locally encoded audio into MP3 files in the browser.
github.com/Vanilagy/mediabunnyLAME MP3 Encoder
LGPL-2.0-or-laterEncodes recordings as MP3 through Mediabunny’s bundled WebAssembly worker.
lame.sourceforge.ioPDF.js
Apache-2.0Reads the text out of PDF files for the anonymizer in the browser.
mozilla.github.io/pdf.js/Transformers.js
Apache-2.0Runs the smart-detection model for the anonymizer in the browser.
github.com/huggingface/transformers.jsDistilBERT multilingual NER (Davlan)
AFL-3.0Finds names, organizations, and places for the anonymizer.
huggingface.co/Davlan/distilbert-base-multilingual-cased-ner-hrlNoto Sans & Noto Serif
SIL OFL 1.1The typefaces used across Weeve Tools.
fonts.google.com/noto