kapre VS matchering

Consider Keras Audio Preprocessors. You can basically use that to insert a convolutional layer into the beginning of your TensorFlow model that has fixed weights that will convert the raw audio frames to STFT and Mel-Frequency Spectrograms.

Songmastr is a web-based AI mastering tool. Utilizing the power of the open-source Python library called Matchering, Songmastr is able to create a masterful audio track that matches a reference song of your choosing. The algorithm studies the RMS, FR, peak amplitude and stereo width of your reference track before applying it to the target audio file.

Ever tried matching ? I really dig that tool: https://github.com/sergree/matchering

I made an automatic mastering website (www.songmastr.com) using the open source software Matchering (all credit to them). I have no users for the meantime, so I'd be happy to get some feedback ! How it works: 1 - You upload your song 2 - You chose a reference for mastering from the catalog / or you upload your own 3 - That's it ! Download the result. The matchering algorithm tries to match frequency response, RMS, peak amplitude and stereo width. The goal is not to replace professional mastering, but rather to offer a quick way to get your demo, or home recording in the same ballpark as commercial songs. Is it free ? - The link below gives out 5000 credits to the first users to claim them - After that, I plan on offering 2-3 free masters per week. Users that need more intensive server use can purchase credits Try it out: https://www.songmastr.com/?oip=eNoztTAx0zM1MjDRs7C0NNAzMjc2BgAmiwPW Have a nice day,

Re: Matchering (https://github.com/sergree/matchering), here is a little more information. If I recall correctly there is a docker-ized version so you can run it locally relatively easily if you are willing to learn a couple of bash commands. I have not played with it a lot and it is separate from the DAW.