I am a bit late for the party - I am just getting into the importing content from YouTube [for French]. I have several questions that I’d really appreciate clarifications on:
I sporadically see videos imported from YouTube available as public (i.e not private) lessons here in the library. But when I use the Chrome extension to import, the videos seem to be private (for me). In fact, if I go to a LingQ course collection of videos imported by me, via the Chrome extension, I cannot even make this course public, it’s hardcoded to private. The question is – is this intentional and does it mean that materials from YouTube should in general be private? This is quite different from the contents of something that’s behind a paywall, in this case it’s free for anyone to watch (think creative commons license etc). A clarification would be appreciated!
When I use the Chrome extension to import from YouTube, the resulting lesson contains text and video (seems to be streaming from the original link). The text in general is subpar to what WhisperAI produces, it is likely the closed captions from YouTube (speculation). The media is video+audio combined, so when I play it, my [smartphone, android] screen should not go dim, can’t switch to other apps etc. In rare cases I see LingQ lessons where there’s both a video and separately audio resource to be played via the LingQ player. Question: Is there a trick to also include the audio extracted, so I can play it while the screen is off etc?
So far the best result of importing content from YouTube (in French) has been to obtain a .mp3 audio extract of the video, import it into LingQ as an audio resource to be transcribed by Whisper AI, wait for a number of hours, and then consume it. It takes a bit of work especially if I need to tweak the audio (trim, split or reduce bitrate etc) to fit in the 60mb / X mins criteria. Ideally if I am going through the hassle, and LingQ is spending resources for transcription, it would make sense for me to make the end result public for others to profit too. Clarification needed – would there be legal implication or other consequences for me/the platform if I do this (I am fine to link to the original resource, original authors, need no recognition etc …)
I’ll take this opportunity to give kudos to the team for hooking up WhisperAI, which typically produces excellent results in speech-to-text. Totally usable.