NVivo

While NVivo allows you to work directly with audio or video files, most people find it much easier to work from a transcription. Transcription can be manual (done by a human) or automated (done by a machine). Manual transcription is usually more expensive, but is also usually more accurate, especially if the recording is at all noisy. Automated transcription is getting better all the time, and with a clean noise-free recording you should be able to expect at least 90% accuracy, and likely much better. Note things you may lose in a transcription include any emotion, pauses, inflection, etc. If these are important to your analysis, be sure to also pay attention to the original source files.

If you're asking someone else, human or machine, to perform a transcription, do pay attention to confidentiality. As of December 12, 2022, Rev.com, Otter.ai, and Transcription Heroes have been reviewed by UofC Security and are approved for use with de-identified Level 2 data (PDF). The onus is on the Principal Investigator to ensure the information being recorded is de-identified. This may include warning participants in a recorded data collection session against mentioning their names, workplace, profession, or any information that may reveal their identities.

As of May 2023, transcriptions made with Zoom (Campus license only) are approved for up to Level 3 data (PDF) and transcriptions made with Microsoft Teams are approved for up to Level 4 data (PDF).

If you have questions about transcription tools approved for studies involving human participants, please contact the Conjoint Health Research Ethics Board (CHREB) at chreb@ucalgary.ca or the Conjoint Faculties Research Ethics Board (CFREB) at cfreb@ucalgary.ca.

Here's a short, alphabetized, incomplete list of automated transcription services, along with costs, links to confidentiality statements, and other considerations.

Note: Many, though not all, of theese services leverage OpenAI's Whisper automatic speech recognition (ASR) system. In late 2024, Engadget posted this list of reports that OpenAI's Whisper invents parts of transcriptions — a lot. Please always remember to listen to the original audio file while reading whatever transcription was created to ensure you have an accurate trainscription, and nothing more!

Automated Transcription Services

Provider	Cost	Confidentiality	Platform	Other
aTrain	Free	N/A (your machine)	Windows, Linux	Offline - 57 languages
Buzz	Free (there's also a paid version if you want a prettier interface)	N/A (your machine)	macOS, Windows or Linux	Offline - slightly techie. More than 90 languages
Descript	4 hours / month for $5 USD	Statement	Web, macOS or Windows	23 languages
ebby.co	$15/hour USD ($0.25/minute) (cheaper with bulk purchase)	Statement	Web	100+ languages
MacWhisper	Free	N/A (your machine)	macOS	Offline - 100 languages
Microsoft Teams	Free (included with U of C Teams accounts)	Statement	macOS or Windows	English only
NVivo	$30/hour USD (cheaper if you buy 5 or 10 hours)	Statement	Web	28 languages
Otter.ai	First 300 minutes/month free	Statement	Mobile or Chrome	English only, 20% education discount should you upgrade
Rev.com	$15/hour USD ($0.25/minute)	Statement	Web	English only
Sonix	$10/hour USD (30-mins free trial)	Statement	Web	35+ languages
Temi	$15/hour USD ($0.25/minute)	Statement	Web or mobile	English only
Zoom (institutional license only)	Free (Included with U of C Zoom account).	Statement	macOS or Windows	English only. Must choose this option before recording. More: Windows / Mac

Human Transcription Services

Provider	Cost	Confidentiality	Platform	Other
Rev	$1/minute USD	Statement	Web	Claims 99% accuracy
Transcription Heroes	$1.69/min CAD	Statement (PDF)	Web	Canadian

Rev is also recommended by the Taylor Institute .

Doing it yourself

Provider	Cost	Confidentiality	Platform	Other
Transana	$75 USD student pricing	N/A (your machine)	Windows or macOS
Express Scribe	$80 USD but 50% off ed pricing	N/A (your machine)	Windows or macOS
f4transkript	25,90 € - 50,90 €	N/A (your machine)	Windows/macOS/Linux	Also offers as a service
HyperTRANSCRIBE	$40 USD student pricing	N/A (your machine)	Windows or macOS

If you ever need to compress or extract just audio from a video file, Handbrake is a good option.

If you really want to dive in to a comparison of the accuracy of transcription services, you could start here .

Here's a web tool  (untested by me) that claims to "convert automated transcripts generated by Zoom or Microsoft Teams to NVivo format, with options to make coding easier."