Loreto Parisi

We have found a good alternative using Inria [OSCAR](https://traces1.inria.fr/oscar/).

@nmstoker yes we have used OSCAR to train a BERT based LM - https://github.com/musixmatchresearch/umberto To download it's like for Italian 69GB so you need fiber channel for sure.

we have the same problem using LASER bi-LSTM model with `PyTorch 1.0 / Python 3.7` https://github.com/facebookresearch/LASER

@chenxijun1029 in which version this has been fixed? Thank you.

@marnovo thank you. At this time I have tried the `macOS 10.13.4 / Thunderbolt 3 Akitio Node / Gigabyte Nvidia GeForce GTX 960 / Nvidia Driver 387.10.10.10.30.106 / CUDA 387.178`...

@marnovo thank you. Also as for today `macOS` update I get a new update from the `Nvidia Driver 387.10.10.10.30.107` as well. Are you aware of specific Nvidia Cards that card...

@marnovo thanks for the update. By the way that thread it is still not up to date at Mar 31, 2018. After that we had as you says `macOS v10.13.5`...

I realized that the calculation should be done for each buffer: ``` while (inFile->eof() == 0) { // .... for(int i=0;ipush_back(fSample); } bpmDetect->process((*inFile).getElapsedMS(), *floatVect); // .... } bpm = bpmDetect->winning_bpm;...

I get the same error ``` peaks.min.js:1 Uncaught RangeError: options.width should be a strictly positive integer value. [0] at module.exports.resample (peaks.min.js:1) at Peaks. (peaks.min.js:1) at Peaks.EventEmitter.emit (peaks.min.js:1) at peaks.min.js:1 ```...

@webchi great result! Wonder if this will work un Ubuntu docker... I can see that the `audiowaveform` binary was dynamically linked: ``` audiowaveform: ELF 64-bit LSB pie executable x86-64, version...