Log in

FAQ - Help

Frequently asked questions about Researchvideo

How reliable is Researchvideo’s AI transcription for research use?

Researchvideo uses WhisperX to produce draft transcripts. It can speed up transcription, but the transcript should always be checked against the original recording before research analysis. Quality varies depending on audio quality, background noise, overlapping speech, dialects, speech rate, language switching, and specialist terminology.

Why are sections or longer passages missing from the transcript?

Missing passages can occur when the model has difficulty detecting or segmenting speech, for example because of noise, low volume, overlapping speakers, fast speech, or unclear pronunciation. Long recordings are processed in smaller sections, and difficult parts may sometimes be skipped or handled incorrectly. Timestamp jumps are a useful sign that the transcript should be checked carefully.

How well does the service identify speakers in group interviews?

Speaker separation is technically difficult, especially in group discussions. It works best when people speak one at a time, voices are distinct, and the recording is clear. It becomes less reliable when speakers interrupt each other, talk over one another, have similar voices, or are recorded from different distances. Manual checking is recommended for research use.

Does the service support Finnish, dialects, spoken language, and multilingual data well enough?

WhisperX supports Finnish and multilingual transcription, but quality varies. Dialects, colloquial speech, unfinished words, children’s voices, fast speech, and language switching can reduce accuracy. The model may also normalize spoken language into more standard written language, so extra checking is needed when exact wording or spoken-language features matter.

How can I ensure my interview is done with good enough quality?

Make sure your surroundings is a quiet as possible:

  • Choose a quiet room with soft surfaces (curtains, carpets, furniture reduce echo). Smaller room is better than a bigger one.
  • Turn off background noise: fans, notifications, TVs, etc.
  • Close windows to reduce traffic/wind noise.

If possible, use quality microphone setup. You may borrow equipment for research purposes from infrabooking.jyu.fi.

Instruct the interviewees to speak clearly and loudly, if possible. Do a 30 second test recording and fix any obvious issues.

My media file does not upload or upload takes too much time?

Some files are massive. Also wireless/mobile connection can be unreliable. Possible solutions:

  • You have VPN on. You do not need to use VPN with Researchvideo, and it will make upload speed much slower. Please disable VPN and try again.
  • Cancel the process and try again later
  • Use fiber connection instead of WLAN or mobile connection
  • Make sure the filename does not contain special characters or multiple file extension names. E.g. "recording-of-yhteistyopaivat.mp4" is a good file name whereas "Recording of Yhteistyöpäivät - JKL_121024.avi.mp4" is not
  • Locally encode the file to a smaller/new version by using software like Handbrake

My media file is stuck in encoding process?

The media file (video, audio) might be in an unknown format, or the file could be slightly corrupted. Possible solutions:

  • Just wait. There might be many encoding/transcription/blur/translation operations in queue. Encoding a file then might take until the next day.
  • Cancel the process and try again later
  • Locally encode the file in a different format, like from .wav file to .mp3 file by using software like Handbrake
  • Locally trim the media file (even a one millisecond might be enough) and then save it again and try again
  • Make sure the filename is well formed and contains only one file extension, eg. "myvideofilename.avi"

Can I move a MyJYU recording from My Media to a Project?

No, but admins can. Please please contact support using HelpJYU and tell the details, like the link to media file and name of the project.

My transcription did not appear after waiting, can I make a new one?

There might have been a random and short network or backend server error. You can easily initiate subtitle generation and translation after file upload, it usually helps. But if not, please contact support in HelpJYU.

I have encountered a new problem, not described above?

If none of the above helps, please contact support using HelpJYU and tell the URL (web address) of your project and error message info and attach screenshots and any other info about the issue.