Arc Transcriber Space (model weights) Copyright (c) 2026 Whissle Inc. Licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0) and the Whissle terms in LICENSE; commercial use requires a separate written licence from Whissle Inc. (ksingla@whissle.ai, https://whissle.ai). Third-party components: - nft2/encoder_onnx (the speech encoder and RNNT decoder, exported to ONNX and quantised) contains a modified version of NVIDIA's stt_en_fastconformer_hybrid_large_streaming_multi (https://huggingface.co/nvidia/stt_en_fastconformer_hybrid_large_streaming_multi), licensed under CC-BY-4.0, Copyright (c) NVIDIA Corporation. Changes: the weights were further trained, and the model was exported to ONNX and int8-quantised. - nft2/decoder contains a modified version of the Moonshine Streaming tiny model by Useful Sensors, Inc. (dba Moonshine AI) (https://huggingface.co/moonshine-ai/moonshine-streaming-tiny, https://github.com/moonshine-ai/moonshine), MIT License. Its licence and copyright notice are kept in nft2/decoder/LICENSE. - examples/samples/*.wav are synthetic speech generated with Kokoro-82M (https://huggingface.co/hexgrad/Kokoro-82M, Apache-2.0) from fictional scripts. They contain no recording of a real person. This model is not affiliated with or endorsed by NASA or NVIDIA.