kaixxx / noScribe

Cutting edge AI technology for automated audio transcription. A nice GUI for OpenAIs Whisper and pyannote (speaker identification)
GNU General Public License v3.0
487 stars 101 forks source link

obsolete, sorry … #53

Closed BabyFnord closed 7 months ago

BabyFnord commented 7 months ago

RTFM!

kaixxx commented 7 months ago

Do you have one of the footswitches mentioned in your original post? Do they send a particular key (F4)?

BabyFnord commented 7 months ago

Yes, I have a three-pedal type VEC USB Footpedal INFINITY USB, and running ioreg -w0 -l -p IOUSB prints:

+-o VEC USB Footpedal@00400000  <class AppleUSBDevice, id 0x1000026b9, registered, matched, active, busy 0 (6 ms), retain 14>
        {
          "sessionID" = 348261090111265
          "iManufacturer" = 1
          "bNumConfigurations" = 1
          "idProduct" = 255
          "bcdDevice" = 288
          "Bus Power Available" = 250
          "USB Address" = 4
          "bMaxPacketSize0" = 8
          "iProduct" = 2
          "iSerialNumber" = 0
          "bDeviceClass" = 0
          "Built-In" = No
          "locationID" = 4194304
          "bDeviceSubClass" = 0
          "bcdUSB" = 272
          "USB Product Name" = "VEC USB Footpedal"
          "PortNum" = 4
          "non-removable" = "no"
          "IOCFPlugInTypes" = {"9dc7b780-9ec0-11d4-a54f-000a27052861"="IOUSBFamily.kext/Contents/PlugIns/IOUSBLib.bundle"}
          "bDeviceProtocol" = 0
          "IOUserClientClass" = "IOUSBDeviceUserClientV2"
          "IOPowerManagement" = {"DevicePowerState"=0,"CurrentPowerState"=3,"CapabilityFlags"=65536,"MaxPowerState"=4,"DriverPowerState"=3}
          "kUSBCurrentConfiguration" = 1
          "Device Speed" = 0
          "USB Vendor Name" = "VEC "
          "idVendor" = 1523
          "IOGeneralInterest" = "IOCommand is not serializable"
          "IOClassNameOverride" = "IOUSBDevice"
        }

Checking its output per virtual keyboard, it does not send a particular keypress. I could not find a way to inspect its USB input but am willing to look into it, if someone can show me how to do it.

kaixxx commented 7 months ago

Information about this pedal is sparse. There is however a software for windows which can be used to send keypresses: https://solutioncenter.stenograph.com/en_US/audio/footpedalguides This is what you want, but for mac. You would configure it to send Ctrl + Space, this controls start and stop of the audio in the noScribe editor. A quick search for mac only revealed this problem report: https://discussions.apple.com/thread/253300520?page=2&sortBy=best Not very helpful...

BabyFnord commented 7 months ago

Hmkay. I have access to another Mac with Karabiner installed. Remembering its manufacturer agnostic feature to read keypresses from USB devices, maybe something useful shows up. More on this in a few days …

BabyFnord commented 7 months ago

Unfortunately, according to the event viewer of Karabiner, this footpedal does not send anything keyboard related. I didn't find a method to inspect something like a raw USB stream, too. Skipping segments in both directions by keyboard would prove to be useful, allowing for faster edits.

BabyFnord commented 6 months ago

@kaixxx On a side note to the foot pedal: With macOS, CMD+SPACE defaults to the OS search function (Spotlight search). While it is possible to deactivate this default behaviour for the time being, some other shortcut or key might be a better choice (as we discussed with @gernophil by mail some time ago) … 💁. Two more keys, for forward/backward navigation of segments, would speed up editing substantially. Thanks for considering!

kaixxx commented 6 months ago

With macOS, CMD+SPACE defaults to the OS search function

Good point. We've also discussed a better alternative for macOS, something with SPACE also. But I forgot... Di you remember?

What would be your usecase for navigating through segments quickly? Should the audio keep playing or not? Will consider this.

BabyFnord commented 6 months ago

Di you remember?

At this time, I can't seem to find that segment of our mail conversation.

What would be your usecase for navigating through segments quickly?

Editing long transcripts faster by using keyboard/shortcut based skipping in either direction.

Should the audio keep playing or not?

If at all possible, depending on whether the audio was playing before invoking the shortcut. If it wasn't, then that would result in even faster skipping segments. I do understand, that there likely be some lag/latency before audio playback starts, hence, I would not expect noScribe Editor to skip through a transcript like a media player would. At least not on my trailing edge machines 😄.

The rationale to this skipping feature is, that it acts as a visual aid for skimming text—if we'd find a dubious word or combination of words in an active segment, the playhead would be already there if we decide activate audio replay.

hanneshier commented 3 months ago

@kaixxx On a side note to the foot pedal: With macOS, CMD+SPACE defaults to the OS search function (Spotlight search). While it is possible to deactivate this default behaviour for the time being, some other shortcut or key might be a better choice (as we discussed with @gernophil by mail some time ago) … 💁.

A quick and easy workaround in the meantime: you can set custom keyboard shortcuts for every menu title in an app via the macOS settings. Just go Settings > Keyboard > Keyboard Shortcuts > App Shortcuts. There, you can add a custom shortcut and limit it to certain apps. You just need to enter the correct menu title, "Play/Pause Audio" in this instance, and set a shortcut (hint: OPT + SPACE was not working for some reason).