What I have:
A kubernetes container, running Ubuntu and JupyterLab; JupyterLab is accessed remotely from a Windows laptop running a Chrome browser
Audio files: Wav format, 16 KHz mono
Various technical derivatives of the audio files. The most easily recognizable to a broad audience is likely the spectrogram, which is: A 2d array of data, where time is displayed on the x-axis and various frequency data are displayed on the y-axis. Each position on the horizontal axis represents an acoustic frame of data, typically high tens or low hundreds of samples. These frames may overlap. This is true of all the 2d derivatives I want to investigate.
(It should not matter, but for the curious, the other derivatives include things like MFCC representations, Mel Spectrograms, MGCEPs, various vocoder outputs, and a few experimental ones.)
What I want:
In general: A tool set (likely but not necessarily something in the vein of a python library centered around widgets) that will, in JupyterLab, allow me to display the audio or audio derivatives, play them, and provide a live visual indicator of what part of the audio is playing.
Most important: The play/display of the raw audio (shown in the link below)
Second most important: The play/display of an actual spectrogram
Ideally: The ability to substitute any spectrogram-like display for the spectrogram.
Example is a typical Audacity display, shown in this video, but I know of no comparable tool in the intersection of python/widgets/jupyterLab.
Note: As the author and end-user of these alternate 2d forms of acoustic data, assume I have detailed (indeed, perfect) understanding of the frame overlaps, frame spacing, etc.
What I currently use:
IPython.display.audio, which plays live audio with a sort of a progress bar (see this link, about halfway down in section 8 for an example; in fact, see just above it in section 7 for an example of a spectrogram) but no real visual display of the data itself.
A sad trombone, because that's just not enough.