Cloudinary Blog

With automatic video subtitles, silence speaks volumes

Auto Generate Subtitles Based on Video Transcript

The last time you scrolled through the feed on your favorite social site, chances are that some videos caught your attention, and chances are, they were playing silently.

On the other hand, what was your reaction the last time you opened a web page and a video unexpectedly began playing with sound? If you are anything like me, the first thing you did was to quickly hunt for the fastest way to pause the video, mute the sound, or close the page entirely, especially if you were in a public place at the time.

If you identify with these scenarios, you are far from alone. A huge proportion of the viewers on social sites and other media-heavy platforms choose to view video without sound. In fact, 2016 studies show that on Facebook, around 85% of video was viewed with the sound off.

But when you are the developer of a website or mobile app with lots of user-generated video content, the consumer expectation for silent video becomes a challenge. All your app users who want to upload their videos of recipes, art projects, makeup tips, travel recommendations, or how to...[anything] are generally very reliant on accompanying explanations to capture and keep attention.

The solution? Subtitles, of-course. Even better? Auto generated subtitles!

Cloudinary, the leader in end-to-end image and video media management, has released the Google AI Video Transcription Add-on, so you can easily offer automatically generated video subtitles for your users' (or your own) videos.

Google AI Video Transcription Add-on registration

Subtitles can speak louder than words

When people scroll through posts or search results with multiple autoplay videos, a particular video has only a second or two to capture viewers' attention. And since the video creators can't rely on sound in most cases, it's almost mandatory to provide text captions to get their viewers interested and to keep them watching, and maybe even to get them interested enough to click on the video and watch (with or without sound) till the end.

The Video Transcription add-on lets you request automatic voice transcription upon upload of any video (or for any video already in your account). The request returns a file containing the full transcript of your video, exactly aligned to the timings of each spoken word.

The transcript is generated using Google's Cloud Speech API, which applies their continuously advancing artificial intelligence algorithms to maximize the quality of the speech recognition results.

When you deliver the video, you can automatically include its transcript in the form of subtitles.

The upload request

To request the transcript of a video upon upload (once you've registered for the transcription add-on), just set the raw_convert upload parameter to google_speech. Since it can sometimes take a while to get the transcript back from Google, you may also want to add a notification_url to the request, so you can programmatically check when it's ready:

Ruby:
Cloudinary::Uploader.upload("lincoln.mp4", 
   :resource_type => :video, :public_id =>"lincoln",  
   :notification_url => "https://requestb.in/abcd123yz", 
   :raw_convert => "google_speech")
PHP:
\Cloudinary\Uploader::upload("my_video.mp4", 
    array("resource_type" => "video", "public_id" => "lincoln",
    "notification_url" => "https://requestb.in/abcd123yz", 
    "raw_convert" => "google_speech"));
Python:
cloudinary.uploader.upload("my_video.mp4", 
    resource_type = "video", "public_id" = "lincoln", 
    "notification_url" = "https://requestb.in/abcd123yz", 
    "raw_convert" = "google_speech")
Node.js:
cloudinary.v2.uploader.upload("my_video.mp4",
   { resource_type: "video", public_id: "lincoln", 
    notification_url: "https://requestb.in/abcd123yz", 
    raw_convert: "google_speech" },
    function(error, result) { console.log(result); });
Java:
cloudinary.uploader().upload("my_video.mp4",
    ObjectUtils.asMap("resource_type", "video", "public_id", "lincoln",
    "notification_url", "https://requestb.in/abcd123yz", 
    "raw_convert", "google_speech" ));
.Net:
var uploadParams = new VideoUploadParams()
{
  File = new FileDescription(@"dog.mp4"),
  ResourceType = "video",  
  PublicID = "lincoln",
  NotificationUrl = "https://requestb.in/abcd123yz",
  RawConvert = "google_speech" )); 
};
var uploadResult = cloudinary.Upload(uploadParams);

The delivery URL

Once you've verified that the raw .transcript file has been generated, you can deliver your video with the subtitles. Just add a subtitles overlay with the transcript file name. (It has the same public ID as the video, but with a .transcript extension.)

If you want to get a little fancier, you can also customize the text color, outline color, and display location (gravity) for the subtitles:

Don't stop there. Show 'em what you got!

Subtitles are great, but if you've already got the transcript file, why not parse it to generate an HTML version that you can show on your web page? This is great for making video content more skimmable and SEO-friendly.

Here's a really simple Ruby script that does exactly that:

require 'json'

class TranscriptUtil

#function receives the transcript input file, path of HTML output file,
#max words to include per timestamp (default 40), 
#and line break for longer entries (default 10).
def convert(transcript_file, html_file, max_words=40, break_counter=10)


    # read and parse the transcript file
    file =  File.read(transcript_file)
    data = JSON.parse(file)


    index = 0
    elements = []
    elementIndex = 0
    words_count = 0
    start_col = "</br><td>"
    end_col ="</td>"

    elements[elementIndex] = "<tr>"
    data.map do |d|
      d['words'].map do |group|
        if(index % max_words == 0)
          elementIndex += 1

#define the timestamp string format          Time.at(group['start_time']).utc.strftime("%H:%M:%S")
          start_time = "      --- " + Time.at(group['start_time']).utc.strftime("%H:%M:%S").to_s + " ---"

#build the html content
          elements[elementIndex] = "<br>" + start_col + start_time + end_col + start_col
          words_count = 0    
        end
        if(words_count == break_counter)
          elements[elementIndex] += end_col + start_col
          words_count = 0
        end
        elements[elementIndex] += group['word'].to_s.strip + " "        
        index += 1
        words_count += 1
      end
    end
    elements[elementIndex+1] = "</tr>"

#save the html content in a new html file
    File.new(html_file, "w+")
    File.open(html_file, "w+") do |f|
      f.puts(elements)
    end
  end
end

You can run the script with the following command:

ruby -r "./transcript_to_html.rb" -e "TranscriptUtil.new.convert('lincoln.transcript','./lincoln_transcript.html',20,10)"

This very simple script outputs basic HTML that looks like this:

sample generated subtitle HTML output

Of course for a production version, I'm sure you'd generate something that looks much nicer. We'll leave the creative design to you.

But if you really want the video content to 'sync' in...

If you are feeling particularly adventurous, you can even add synchronization capabilities between the textual display of the transcript and the video player, so that your viewers can skim the text and jump to the point in the video that most interests them. You also sync the other way, making the displayed text scroll as the video plays, and even highlight the currently playing excerpt.

Demonstrating these capabilities is beyond the scope of this post, but we challenge you to try it yourself! We've given you everything you need:

The Cloudinary Video Player can capture events and trigger operations on a video. Use the player in conjunction with the Google-powered AI Video Transcription Add-on, add a bit of javascript magic, and you'll be on your way to an impressive synchronized transcript viewer on par with YouTube and other big players in the video scene.

The bottom line (or should I say, "The subtitle…" ;-)

As it becomes more and more commonplace to use videos as a way to share information and experiences, the competition to win viewers' attention becomes increasingly tough. The Video Transcription add-on is a great way to offer your users automatic subtitles for their uploaded videos, so they can grab their audience's attention as soon as their silent video begins to autoplay. Oh, and it's great for podcasts too.

To watch it in action, jump over to the Cloudinary Video Transcoding Demo. Select one of the sample videos or upload your own, and then scroll down to the Auto Transcription section to see the transcription results. And while you're there, check out the many cool video transformation examples as well as a demonstration of the Cloudinary Video Tagging add-on.

But the real fun is in trying it yourself! If you don't have a Cloudinary account yet, sign up for free. We'd love to see your demos of the sync implementation suggested above. Please add a link in the comments to show off your results!

Recent Blog Posts

Reimaging DAM--The Next-Gen DAM for Marketing & Dev

There are great digital asset management (DAM) products out there for uploading, storing, managing, organizing, and sharing digital assets. With Cloudinary's new end-to-end DAM solution, you can also upload and manage your assets efficiently, but the journey doesn't end there. It continues on to the development and delivery stages, so that your assets can be seamlessly manipulated, optimized, and delivered to create an engaging user experience that will in turn, increase conversion and loyalty.

Read more
Integrating Cloudinary with Forestry’s Media Library

At Forestry, we believe that there is a bright future for static HTML sites built with tools like Jekyll and Hugo. These tools can create sites that run well, and are easy to host and maintain, because they don’t require any server-side code.

Read more
Video Optimization With the HTML5 <video> Player

Lack of experience and compression knowhow can cause significant user-experience problems. For instance, on a major retail site, I recently ran into a 48 MB video-hero banner. Pulling out the video and encoding it as an H.264 MP4 reduces the size to 1.9 MB. So, despite the desire for more video content, developers have not yet caught up to best practices. How do we get the best of both worlds without creating a disaster like the one above?

Read more
Build a Facial Emotion Recognition Based Video Suggestion App

Developers are always looking for new and creative ways to deliver content that resonates with the way users feel. Often using the latest technical innovations the market has to offer such as Artificial Intelligence (AI) and Machine Learning (ML). What better way to demonstrate innovative uses of these technology in a consumer market than capturing expressions from your users and then serving content based on that expression!

Read more
Improve Customer Data Protection with GDPR Implementation

TL;DR

Yay! We've done it! Gold-Star for us! We've talked with all the people, made all the changes, paid all the lawyers and checked all the boxes. GDPR? ✅Done!

Not so fast. Of course, conforming to the GDPR regulations introduced in Europe is just the beginning. This is a process and a state of mind that must become part of our long-term cultural ethos.

Read more