Creative Communities of the World Forums

The peer to peer support community for media production professionals.

Activity Forums A.I. Why do all auto-captioning softwares seem to be online ? 😱

  • Why do all auto-captioning softwares seem to be online ? 😱

    Posted by Jacob Giacometti on January 9, 2023 at 11:45 am

    Bonjour !

    I was trying to find a kind of plugin, or even standalone, simply to create some auto-captioning like it is very commonly done around social media nowadays : the software analyses the audio, converts it to text, and aligns it to the audio giving you the option to select how many words per “slide” you want. Very common thing nowadays !

    But has anyone else also noticed that this technology seems to only exist in the form of “online” versions, with paid plans ? I was wondering how come there aren’t as well software/plugin versions, one could download to his computer and use his own microprocessor to do the processing !

    Very curious on your thoughts about this !

    Anni Gerard replied 8 months ago 9 Members · 8 Replies
  • 8 Replies
  • Mads Nybo jørgensen

    January 9, 2023 at 12:21 pm

    Hey Jacob,

    Good Point.
    Adobe Premiere Pro will off-line let you create Speech to Text, with time-code (even just for transcript, I use it on most productions).
    If you want something without a subscription, there is also Nuance Dragon which will allow you to import audio files (I only checked the Professional version).

    However, reason for why most services are on-line, is because the language libraries they access keeps on improving and expanding – I understand that both Goggle and AWS runs language libraries, just to mention the “big” ones.

    I am a fan, and user of Nova, which does a lot more, including logging your rushes, than just transcribing and creating subtitles. Because of what is under the hood, they can also translate the transcript from any selected language, to a large number of languages and create subtitles (although not 100% accurate, sometimes as low as 70% – depending on source and/or target language). Over the last year I have for my clients transcribed English productions and translated to Spanish (Mexican), and a few African languages.

    Don’t know if my link will give you a special offer, but do feel free to take it for a spin:
    https://app.wearenova.ai/signup?ref=madsnybojorgensen2
    Although, this is an on-line service only.

    Hope that this helps.

    Atb
    Mads

  • John Fishback

    January 9, 2023 at 10:40 pm

    There’s an app for Final Cut Pro called Captionator on the App Store, which I believe uses a local library. It’s a bit of a pain to format the captions, but its accuracy is pretty good.

  • Ben Balser

    February 24, 2023 at 9:34 pm

    Scribeomatic and SymonSays are both extensions for FCPX and work really great.

  • David Norden

    March 16, 2023 at 2:58 pm

    Hi Jacob, and everyone else. We are soon to roll out tool that will read Avid project files/mxf-files and auto-transcribe them (individual or batches). The beta will launch in a week or two, and you can read about the apps on our site, http://www.edmon.ai . Once completed, our apps will allow for transcribing a batch of clips from a Media Composer bin, edit the parts you select (via text-editing), exporting an AAF that links to the original media in your Media Composer project. Let me know if you would be interested in trying it out (windows only, for now). We will soon add subtitle-export as well – which is really what you asked for I guess 🙂

  • Cole Mendoza

    April 11, 2023 at 2:07 pm

    Online auto-captioning tools are more common because they can use cloud-based resources, which are cheaper and more scalable. But there are also downloadable tools like Amara and CaptionMaker that can do auto-captioning offline.

  • Ernest Rosado

    May 3, 2023 at 3:56 pm

    The best transcription engines (the ones that rely on AI rather than traditional speech recognition) are extremely computationally expensive. OpenAI’s Whisper is one of the best and it’s open source, but it’s heavy and slow, slower than realtime, especially on consumer hardware. You could run it yourself and let it do your transcriptions, but it’s definitely not a one-click plugin. You’ll need to do some scripting to get it into a format you can use.

    All this will improve over time but for now the AI stuff really runs best on enterprise hardware, which is why most of these are cloud services.

    The engines also need updates. New words, phrases and names enter the lexicon all the time and if they are to be detected, the models must be maintained, which is a pretty good argument for a subscription business model.

  • Craig Seeman

    August 28, 2023 at 4:27 am

    I’ve used MacWhisper and now Ulti.Media Transcriber which is based on MacWhisper. It does depend on the “engine” but Transcriber uses the “Large” model and it seems to be real-time or better for me on my Mac Studio M1 Max. It would be slow on an Intel Mac though.

  • Anni Gerard

    September 25, 2023 at 11:48 pm

    I use an auto subtitle software and it’s online too.
    I’ve used other offline software but to me they felt weirdly inadequate and vastly inaccurate at times. Their bug fixes were slow and modifications were ultra painfully time-consuming hence switched to the online ones

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy