January 15, 2021
Recollective: 2020 In Review
Before we look ahead at 2021, it's time to recap all the great Recollective product improvements we shipped last year such as our new Live Activities.
As of today, all Recollective sites have received a free and significant new feature: Fully automated video transcription.
No longer will your participants' most vivid and candid insights be trapped inside a video everyone must watch. Every single word uttered in a video will now be right there on the page for you to read, search, excerpt and summarize.
With support for over 110 languages and variants, our ever-improving AI-powered speech-to-text conversion will accurately transcribe all the spoken audio content in participant-submitted videos. This includes webcam recordings, mobile device submissions and traditional video uploads.
Once your site has the new software update, the new transcription feature will be active. A text transcript will appear directly below or alongside participant-submitted videos so you can listen and read simultaneously.
We are already working on an interactive transcript experience. Recollective will soon highlight words as they're spoken and allow word selection to control video playback.
UPDATE (Oct 3, 2017): Recollective now has fully interactive video transcripts!
Machine learning and artificial intelligence are amazing as they get better with time. Accuracy can still vary based on sound quality and accent variations. Recollective provides you full control to improve the transcripts that are generated to ensure your word clouds and text excerpts are accurate.
First, it's important to know that such transcripts are a raw extraction of text and won't include some of the things you might expect, like punctuation.
If you ever receive a transcript that is of very poor quality, we recommend you check the audio quality of the video during its playback. If it's good, ensure the language shown in the transcript header matches the actual language spoken in the video.
Recollective offers an option to instantly re-transcribe a video with another language or language variant to improve accuracy. For example, we support 13 English and 20 Spanish accents and so selecting the closest match will greatly increase the transcription accuracy.
Once you're satisfied with the automated transcript, you're still welcome to make minor edits directly to the transcript. To do so, select "Edit" from the menu above the transcript.
UPDATE (Oct 3, 2017): You can now just double-click any word to instantly edit or remove it. This update is part of the new interactive video transcript. There is no longer a dedicated edit mode.
If you have a study already underway, you might be wondering if Recollective will transcribe the videos you've already collected. The short answer is, yes, please contact us if you are conducting an active study with older vidoes that need transcription.
Recollective can separate the audio from all past submitted videos so the transcription engine can convert the spoken words into text.
Valuable insights found in your video transcripts can now be saved and easily reported. Just highlight one or more words to reveal the "Save Excerpt" menu option. This follows the same process available for saving verbatim excerpts on all other text submitted by participants.
Once a snippet of text is saved, you can apply one or more codes to the excerpt. This means all video content can finally be included in your Excerpt Reports.
In forthcoming releases of Recollective, excerpts from video transcripts will automatically generate a matching video clip! You'll be able to quickly assemble different "highlight reels" without bulky video exports or complex post-processing!
If you are not content with the transcripts generated by this new feature, be aware that the transcription feature can be disabled for your site. Simply contact us via our standard support channels to share with us any concerns you have. We're always looking to improve.
Recollective's interface is fully translated into 18 unique languages and, in this new update, we can now recognize over 110 spoken languages and variants.
UPDATE (June 2020): We now support 24 languages for the interface and 127 spoken languages and variants. View our resource article on Supported Languages.
Below is the complete list of the spoken languages supported for video transcription. In our next release, we'll allow participants to select their spoken language variant prior to video submission to maximize transcription accuracy.
We updated this list in April 2020 to also include:
We look forward to your feedback!