I have put together the attached PPT as a personal reflection about what I have learnt about the use of AI to transcribe video interviews in the Covid 19 world. In the PPT I cover questions such as:
1) why transcribe when you have the audio visual?
2) what types of AI transcribing are available - a tip, there is a difference between Dictation and AI transcribing
3) provide some links to a few transcribing sites, offer some quick reflections on them, their costs and service structure
3) a walk through of Sonix, which is the AI service I have been using.
2. There are two parts to this PPT
1. 1. Introduction to what computer aided
transcribing is, why do it, who should do it, an
example, and costs
2. 2. Walk through using SONIX to AI transcribe
a video/audio file, and editing it
3. 1. Part1comprising:
Whytranscribe
qualitativedata
Who shoulddoit: A
personalview
WhatisAItranscribing
An example ofAI
transcribing
Costs
Why transcribe? A personal view
Key part of the qualitative analytical process
that involves active listening and thinking about
what is being said, how it is being said and who is
saying it
Record of data collection activity
Enables use of text in a variety of ways for
analysis and reporting
Why AITranscribing ?
Efficiency and Effectiveness
Combines elements of visual with audio – best of
both worlds?
4. Why transcribe
focus groups /
in-depth
interviews, and
who should do it: A
personal view
Who should transcribe?
Anyone who claims to be a
qualitative researcher/evaluator
it is a way to keep grounded.
Inherently, qualitative research is
grounded in social behaviour –
you have to see it, feel it, listen
and talk it, draw and write it
5. Two types of
computer
aided
transcribing
activities
DictationTranscribing
This is where you simultaneously listen to the
audio/video and speak it out aloud (dictation) and
the software transcribes what you are speaking.
Both MSWord and Google Docs offers this, and it
does work.
Here are some HowTo video clips
https://collegeofthedesert.instructure.com/course
s/11025/pages/how-to-quickly-convert-audio-to-
text-free-and-easy-using-google-chrome-dot
https://qz.com/work/1087765/how-to-transcribe-
audio-fast-and-for-free-using-google-docs-voice-
typing/
6. Two types of
transcribing
activities
Thinking points for Dictation
Slow your speaking down
Means you need an audio/video player that
allows you to slow the playback down in various
increments to reflect your listening, typing, and
speaking. So normalWindows player isn’t
suitable.
You need to manually:
organise your paragraphing / sentencing as you
go
inset time stamps
Time stamps enable the researcher to quickly and
easily locate the audio/text of interest.
This is where the audio cutting software comes
into play as instead of cutting text, you cut the
audio and organise much as you would text in
NVIVO.
7. Two types of
transcribing
activities
Artificial Intelligence (AI)
Transcribing
AI based transcription services are
essentially all web based (cloud computing)
You upload your video or audio file to the
service, which then transcribes it, and you
download the transcript as aTXT, Docx, or
PDF file for editing.You can edit on-line.
8. Two types of
transcribing
activities
Thinking Points for AITranscribing
Need a fast and stable internet connection
Accuracy of the transcription materially reflects the quality
of the audio
Where the audio is clear, it is surprising good –
allowing for regional accent differences (i.e., the
classic Kiwi / American issue of Auckland becomes
Oakland).
Better services offer
Cutting of video clips
Rich text editing functionality, re-paragraphing, find
and replace
Auto realignment of time stamps
Sharing of links to transcription and video/audio files
Downloads of the original video/audio file in a
compressed format that is playable
9. Examples ofAI
Transcribing
providers
available
Examples of providers:
https://sonix.ai/how-to-convert-mp4-to-text
https://otter.ai/login
https://transcribe.wreally.com/convert-MP4-to-text
https://go-transcribe.com/convert-mp4-to-text
https://www.360converter.com/conversion/video2Te
xtConversion
https://www.happyscribe.com/convert-mp4-to-text
Give the AI transcribing a go, you may be in for a
pleasant surprise.
Next section - using Sonix as an example
11. File is open: Functionality includes
Paragraphing is
automatic and
editable
Adjust the speed
of the play back
Tabs to
settings
Enter the name of
the speaker: Note the
time stamp
Check box signals
completion of
editing
Video playback
synced to text
13. Functionality includes:
Exporting results in various
formats
• note the Underlying Media
gives you a compressed file
version – example, the
original uploaded file was
500 megabytes, now 109
megabytes and opens
automatically in Windows
Player
17. What doesAI
Transcribing
cost?
Costs vary by
provider
type of plan (e.g., personal, corporate,
NGO)
level of service (basic / advanced)
Most charge an hourly rate per audio length
of time
or a base rate plus hourly rate
plus $ for service add-ons – increased
functionality (e.g., allowing multiple
people to edit, auto re-time stamping)
28. HereistheAITranscriptreceivedback:Nowtimetoeditit
Slow the playback down
Label the speaker
You can:
• type and jump around in the text
anywhere
• Re-paragraph by pressing ‘Enter’
key or join two paragraphs by
‘Deleting’ space after last word
• Audio stops automatically when
you type – the ‘Tab’ key stops /
starts the audio playing
Video plays back in
sync with the text
29. This tells me how
confident the AI is
The colour highlight
points me to words the AI
thinks are problematic
31. Downloading the
finished product
- Select Export,
choose format and
other options,
download
note the Underlying Media
gives you a compressed file
version of the video/audio
example, the original
uploaded file was 500
megabytes, now 109
megabytes and opens
automatically in
Windows Player