ldenoue

6 ms·

31.
by ldenoue 1y ago
Full playable transcript https://www.appblit.com/scribe?v=_PioN-CpOP0
32.
by ldenoue 1y ago
Full playable transcript https://www.appblit.com/scribe?v=LCEmiRjPEtQ
33.
by ldenoue 1y ago
How is the turn detection working? LLM prompting or a special AI audio plus text model?
34.
by ldenoue 1y ago
Readable video https://www.appblit.com/scribe?v=0xS68sl2D70
35.
by ldenoue 2y ago
Here’s one video above 1 hour and it works with Scribe https://www.appblit.com/scribe?v=FQUo2r-ow-k
36.
by ldenoue 2y ago
looks great. I made a similar app called Scribe where you can highlight passages of the transcript. It's working on the web but also as an iOS app. https://www.appblit.com/scribe To solve the server IP sometimes being
37.
by ldenoue 2y ago
Same method as my open source lib https://github.com/ldenoue/readabletranscripts but several folks asked for a hosted version so here you go. 200 free minutes on signup so you can try for free. LLM corrected transcript
38.
Show HN: Free minutes for LLM-powered YouTube transcripts(appblit.com)
2 points by ldenoue 2y ago|1 comments
39.
by ldenoue 2y ago
This thing is amazing can you add recording? Perhaps some samples you or visitors create? Then add a little sampler for beat and it’s a fantastic tool
40.
by ldenoue 2y ago
This is the version of my open source library that many people liked https://news.ycombinator.com/item?id=42238890 This will give you 120 free minutes to try out this LLM powered YouTube correction.
41.
Show HN: Scribe Pro = LLM powered YouTube transcripts correction(appblit.com)
2 points by ldenoue 2y ago|1 comments
42.
by ldenoue 2y ago
LLM corrected transcript (using Gemini Flash 8B over the raw YouTube transcript) https://www.appblit.com/scribe?v=YD-9NG1Ke5Y#0
43.
by ldenoue 2y ago
Did you see this using the API or the online product gemini?
44.
by ldenoue 2y ago
It’s not uploaded anywhere: the client fetches Gemini servers directly from your browser. But I understand it can be difficult to trust: that’s why the project is on GitHub so you can run it on your own machine and look at how the key is us
45.
by ldenoue 2y ago
Although Gemini accepts very long input context, I found that sending more than 512 or so words at a time to the LLM for "cleaning up the text" yields hallucinations. That's why I chunk the raw transcript into 512-word chunks
46.
by ldenoue 2y ago
Do you have something to read about your study, experiments? Genuinely interested. Perhaps the prompts can be made to tell the LLM it's specifically handling human speech, not written text?
47.
by ldenoue 2y ago
I should do that, let me try.
48.
by ldenoue 2y ago
The current Flash-8B model I use costs $1 per 500 hours of transcript.
49.
by ldenoue 2y ago
Definitely: and just giving the LLM context before correcting (in this case the title and description of the video, often written by a person) creates much better transcripts.
50.
by ldenoue 2y ago
My repo doesn't re process the audio track: instead it makes the raw ASR text transcript better by feeding it additional info (title and description) and asking the LLM to fix errors. It is not perfect, it'd sometimes replace word
51.
Show HN: Gemini LLM corrects ASR YouTube transcripts(ldenoue.github.io)
177 points by ldenoue 2y ago|109 comments
52.
by ldenoue 2y ago
I also created something web-based last year called ScreenRun https://screenrun.app It really works best in Chromium-based browser which have WebCodecs.
53.
by ldenoue 2y ago
I was playing with Gemini this weekend to solve that issue and your video transcript converted into malt. Here’s the result: Clean YouTube Transcripts appblit.com Created by @ldenoue , this experiment uses the Google Gemini gemini-1.5-flash
54.
by ldenoue 2y ago
I use firebase auth with Google, Facebook and email (magic link) This is live at https://screenrun.app/
55.
Ask HN: Do you need an app to manually clip a video from its transcript?
1 points by ldenoue 2y ago|1 comments
56.
by ldenoue 2y ago
Working on restoring punctuation marks in raw YouTube transcripts. Bert model trained as a token classifier works but it doesn’t scale well with more languages. See https://www.appblit.com/scribe for the current solution No
57.
by ldenoue 2y ago
Same here: I work on speech for a big company and working from the office is terrible. I had to squat the restroom 3 times yesterday so I could work (talk to my phone!) At home I am so much more productive and zero commute. Because of the b
58.
by ldenoue 2y ago
You can also try Scribe (free chrome extension and iOS app) https://www.appblit.com/scribe
59.
Ask HN: Should I open source my SaaS ScreenRun
1 points by ldenoue 2y ago|1 comments
60.
by ldenoue 2y ago
The new ai voices feature (text box below the stage) lets you generate real sounding audio from any text you type. But you can also record your mic and have speech to text recognize the text. In both cases the subtitles appear over your vid

More ›