The AI-Transcription functionality requires a specific license. Contact our sales team if your installation does not offer the commands described in this section.
AI-based Transcriptions can be performed on previously recorded videos, right from within VideoSyncPro Studio. It generates a plain text file name *.transcript which holds only text without time information. You can also simultaneously create an *.srt or *.vtt sub-title file including the required time information.
IMPORTANT: All AI-based transcriptions are a great help but not perfect. We are not responsible for any misinterpretations by the AI and all automated transcriptions must be verified by a human.
Verify Settings
The settings for the Whisper generated auto-transcriptions in VideoSyncPro Studio are available under Settings > General:
▪Click Open AI-Transcript settings at the bottom of the dialog.
▪To allow the user to make changes before every transcription, activate the option Show settings before transcription.
▪Select your preferred Model from the drop-down list.
Note: What Model fits your work best is something you need to try out. 'Tiny' and 'small' models are much faster than the bigger ones, but not very good.
'Base' is quite good for starters and runs on not so powerful computers. 'Turbo' is much better.
The Large models require much more GPU power.
For English language there are separate *_en models trained for English only.
▪If your recording contains more than one speaker and speaker voices are easy to tell apart, you can activate Perform speaker recognition and specify the number of speakers.
Note: Speaker recognition won't be perfect and usually takes up more time than the transcription itself, so only activate this option if you really need it.
▪Select the required transcription output file format under Transcription format:
oNone - No transcript is created oCSV - The resulitng csv file contains time stamps and transcriptions. oSRT - The resulitng text file contains time stamps and transcriptions in the official SRT subtitle format. oVTT - The resulitng text file contains time stamps and transcriptions in the official VTT subtitle format. oProtocoll - The resulting text file contains transcriptions grouped by speaker. |
oUnder Transcript type you can choose between 'Per sentence' and 'Per Word'.
Per default, the AI will always try to produce grammatically correct sentences, which is good when you are mainly interested in the content of the discussion.
To keep hesitations, word repetitions and such:
▪Activate the option Verbatim transcription (include hesitations and mistakes).
IMPORTANT: All AI-based transcriptions are a great help but not perfect. We are not responsible for any misinterpretations by the AI and all automated transcriptions must be verified by a human.
oThe option Highlight words on subtitles prepares an *.srt so that a compatible player can highlight the words accordingly.
There is currently no sub-title function available in VideoSyncPro Studio.
▪Select your preferred output path method: Same as Audio/Video file automatically stores the output file inside the recording folder of the selected recording.
**) The GPU calculation is faster if your GPU supports CUDA.
Start AI-based Transcription
If the settings match your future transcriptions, you can run the routine from the Home screen:
▪On the Home screen, make a right-click on the recording you want to transcribe.
▪Select AI transcript from the context menu:
▪Select a source based on which the transcription should be made.
Note: Make sure the audio quality is very good. Select an Audio source if possible, because the quality is usually better than that in the combined A/V file.
Depending on your settings, the AI-settings dialog explained at the beginning of this topic will appear, allowing for last minute changes.
▪Click OK to start the transcription based on the predefined settings.
The progress is visible in the status line at the bottom of the screen:
When the transcription is finished, you receive a notification.
The transcription files are created in the specified folder.
TAKE CARE: Running the transcription routine again with other settings will overwrite any previous transcription, so move or rename the created file to keep it.
Transcription Result
The auto-generated AI-transcription results look like this:
Transcription Protocol
The Transcript format 'Protocol' generates a *.transcript file which can be opened in any text editor application. This file contains the complete transcription but does not hold any time references:
Subtitle File
If you specified an other Transcript format file type, you'll find the resulting *.srt, *.vtt or *.csv file in the specified folder.
Subtitle Content
*.srt
|
.vtt
|
You can use the subtitle format *.srt and *.vtt files for:
oImporting into Mangold INTERACT for further analysis and be combined with visual observation Events.
oAs subtitles when playing your video(s) in a compatible player like the VLC-Player.
oAdding subtitles to your video on Youtube.
The *.csv file can be used in Excel and other applications.
