-
Art doesn’t pay
-
[Michael Gissing] “Given the number of X seats in the world, even if all 30,000 of the Facebook group respond, it may say more about the demographics of that group than how X is really being used as it’s a tiny percentage of the millions of X installs”
Yeah, it might be a big enough sample if some real data scientists were digging in and both designing the poll and running the numbers, but it’s just an opt-in quick poll of those motivated to respond, and as such, definitely questionable for more than just idle conversation.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
[Bill Davis] “In their niche, VERY powerful. However, their niche isn’t particularly large, IMO.”
I guess I just don’t see it as being a niche-type solution. I’ve done everything from scripted to reality TV to weddings to corporate to documentaries to live event coverage and I can see transcript-linked-video being potentially useful in just about any scenario that involves someone speaking on camera.
[Bill Davis] “So transcriptions will become just another software provided expectation. “
I think this is more the crux of the issue. Today there are technological and monetary hurdles that make transcripts (especially at the level of functionality Builder offers) an additional cost so each person in each scenario has to decide if that additional cost is worth it to them. Now imagine a day when those technological and monetary hurdles are removed from the equation and every NLE generates transcripts, links them to media, and allows for the kind of text-based editing and marking up that Builder provides as a matter of course. I think people will quickly latch onto it and come up with ways to utilize transcripts and text-based editing and markup that they never even thought was possible before.
-
[Andrew Kimery] ” Today there are technological and monetary hurdles that make transcripts (especially at the level of functionality Builder offers) an additional cost so each person in each scenario has to decide if that additional cost is worth it to them. “
It’s also the time involved. You have to get that footage to the people who will transcribe it and that means transcode the footage and then upload time, and for a doc with hours and hours of footage you are waiting for a while. If you have the time to wait………
-
[Tony West] “It’s also the time involved. You have to get that footage to the people who will transcribe it and that means transcode the footage and then upload time, and for a doc with hours and hours of footage you are waiting for a while. If you have the time to wait………”
That’s how it used to be. With the incredible advances in speech recognition software, transcripts literally only take minutes to create and that’s why they are even more powerful today.
Windows 10 Pro | i7-5820k CPU | 64 gigs RAM | NvidiaGeForceGTX970 | Blackmagic Decklink 4k Mini Monitor |
Adobe CC 2018 |Renders/cache: Samsung SSD 950 Pro x2 in Raid 0 | Media: Samsung SSD 960 PRO PCIe NVMe M.2 2280 | Media: OWC Thunderbay 4 x 2 Raid 0 mirrored with FreeFileSync -
[greg janza] “That’s how it used to be. With the incredible advances in speech recognition software, transcripts literally only take minutes to create and that’s why they are even more powerful today.”
Which software are you using Greg to do it? I have not had good experiences with speech software. Too many mistakes. I recently used Rev for German subtitles for my film because they use people to transcribe and so far I have been happy with the results, but they only accept certain formats to upload so unless your source material was shot on that, you would need to transcode first then upload. That’s the time I was talking about.
-
[Andrew Kimery] “I’ve done everything from scripted to reality TV to weddings to corporate to documentaries to live event coverage and I can see transcript-linked-video being potentially useful in just about any scenario that involves someone speaking on camera.”
Wow, that’s truly interesting.
If you don’t mind me asking, what type of transcripts have you pulled from a wedding? Would it be for a best man speech type thing? Or perhaps something like a reception camera set up to do automated talking head best wishes? I’m having a bit of difficulty understanding how a transcript would help in an event type of situation.
I know modern weddings can be very big shoots at the high end, but I know little about them.
So has transcription use in modern event work become a normal thing?
Just curious.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
Rev does a very good job with a relatively quick turnaround time but traditional transcription services just don’t have much of a future.
Digital Anarchy makes a Premiere plugin called Transcriptive. The plugin works in partnership with the two major online speech recognition engines – Speechmatics and Watson.
The plugin bounces any interview to one of the two engines, the speech recognition engine does the heavy lifting of analyzing the audio and creating the transcription and then that file gets bounced back into the plugin interface in Premiere and gets married to the media file.
There are caveats. It costs money to use the speech engine services and the speech analysis makes mistakes. But here’s why it’s the future of transcription, the mistakes are minimal. I’d guess that it’s accuracy rate is about 95%. The cost of using the speech recognition services is a small fraction of the cost of any traditional service and most importantly the turnaround time for this is a few minutes. For my most recent project I had several 30min interviews and I was able to get transcripts created in less than five minutes.
The only limiting factor at the moment is that it’s a Premiere plugin only but I’m sure that will change.
Windows 10 Pro | i7-5820k CPU | 64 gigs RAM | NvidiaGeForceGTX970 | Blackmagic Decklink 4k Mini Monitor |
Adobe CC 2018 |Renders/cache: Samsung SSD 950 Pro x2 in Raid 0 | Media: Samsung SSD 960 PRO PCIe NVMe M.2 2280 | Media: OWC Thunderbay 4 x 2 Raid 0 mirrored with FreeFileSync -
[Bill Davis] “If you don’t mind me asking, what type of transcripts have you pulled from a wedding? Would it be for a best man speech type thing? Or perhaps something like a reception camera set up to do automated talking head best wishes? I’m having a bit of difficulty understanding how a transcript would help in an event type of situation. “
I haven’t done transcripts for a wedding before, but your example is the sort of thing that I would use them for. In the wedding videos that I did in the past I’d cut a 10min or so highlight reel and then do very loose editing on the rest of the event to get it down to around 90 minutes. If the client wanted a different sound bite from the Best Man (the joke about the horses in the middle of his speech, not the joke about the cars towards the end of his speech) or the inclusion of dad mentioning Aunt Jayne (which didn’t stand out to me as significant, but apparently the bride is super close to Aunt Jayne) then having an transcription would allow me to easily search for those specific moments and jump right to them.
Other events can be the same way. For example, I’ve cut hype/sizzle reels for press conferences before and I always have to watch the whole thing (typically 60-90min) in order to find a handful of juicy soundbites. If I had a transcript of the speaker(s) at the event I could skim that much faster than I could go though the footage (even at 1.5x or 2x playback). Recently at work we were tasked with pulling some sound bites of an on-air talent to be used during an awards ceremony. Faced with the daunting task of going through hours of footage looking for great one-liners from said talent, someone had the idea of searching YouTube (this particular talent has a big YouTube following). We found fan-made vids highlighting our talent so we just had to look for air-date clues so we could find the original footage on our server. It was a creative time saver, but certainly not a very widely application solution.
Reply to this Discussion! Login or Sign Up