If your interview is using online(internet chating or using system then use snagit and camtasia studio for recording and editing).

Hi I am audio recording my interviews. The interviews range between 1 to 3 hours each. So it is quite exhausting transcribing

Hi! We use this one: https://www.audiotranskription.de/english

Vey user-friendly.

Thanks Katharina. Is it available free

Couple of ideas.  You might try Dedoose, for uploading audio directly and then transcribing only those parts of the file that you will actually be analyzing within Dedoose.  A separate software just for transcribing that works well is Inqscribe, but it does have a license cost -- well worth it if you have a lot to transcribe.

I'd also suggest f4, http://www.audiotranskription.de/english/f4.htm available freely for Mac OS X and Windows.

express scribe is also a good free software http://www.nch.com.au/scribe/

If you want to to transcribe videos then you need thr Pro version which is not free, but the audio version is 100% free. 

Thanks all for your suggestion. I will try all. I tried using the NCH to transcribe one interview. In the second attempt to transcribe another piece it asked me to purchase. So I wonder if I got the wrong software.

If you have something like Dragon, you can listen to the interview and then recite it into the software in shote chunks. Using software that allows you to slow down and pause the recording will definitely help with this process. 

Even if you "train" the software, you will still have to correct a number of errors that it makes, but it is the only way I know to use this kind of speech recognition software for transcription.

You may omit transcribing completely by analyzing directly the audio files.

http://www.qualitative-research.net/index.php/fqs/article/viewArticle/513/1110

Regards

Stefan

transcriva is what I use, even though it is outdated.

http://otranscribe.com/

also works.

Transcribe is worth at look: https://transcribe.wreally.com

Below are the features I look for in transcription software. I’d welcome hearing if others look for different things:

- skip forward or backward by set amounts (X seconds or Y minutes)

- speed-up or slow down the playback speed (in the range of 0.8x to 1.2x)

- do the above using the keyboard to minimize disrupting your typing

- add timestamps

Text/font formatting options are nice but i can live without that.

Voicebase mobile app and web based is also an interesting option. The computer transcription can be helped by providing subject specific dictionaries. New users are offrered 50 hours of free transcription, which makes this a cost effective solution for smaller projects (e.g. undergraduate research). 

The software does some keyword analysis for you out of the box, and links text and audio, so when searching/reading in the text, it's easy to hear the specific audio section.

Outsourcing transcription (to a computer or a human) has one mayor drawback : the researcher risks to skip over / forgets some materials. yes the transcription process is long, but it forces the researcher to listen again to each word that was recorded.

Declaration: I have no relationship whatsoever with Voicebase

https://www.voicebase.com

please who can give some good transcription softwares for a qualitative study using in depth interviews 

After reading this post and tried many of the links. I found the best transcriber is Google Docs with speech to text option. Works very well and free. You can use it manually or through API https://support.google.com/docs/answer/4492226?hl=en 

https://support.google.com/docs/answer/4492226?hl=en

Thank you for these useful suggestions. I looked at some of these options, and I am surprised that in all the cases I saw, the human still needs to do the transcribing (if I understand correctly).

This surprises me, given the massive improvement in voice recognition recently. Is there no software that would take an audio recording, and produce a text file -- of course with errors that a human would need to correct? 

Many thanks

Jan,

If you dictated an audio file, you could indeed send that to a transcription program and get a reasonably accurate text file. But when you switch to naturalistic settings, such as interviews, and especially when you have more than one speaker, the situation is still too complex for current commercial software.

After reading this post I am going to experiment with Dragon and google.  Is there anything better.  Alternatively, do dictaphones with built in speach recognition software work any better ?

So what I am reading - and based on my own take - there is no simple solution due the complexity of the issue.   

For analysing data I find that ELAN is a good tool.

https://tla.mpi.nl/tools/tla-tools/elan/

From my experience, Computer-generated transcription can provide some support. For example, you can upload your audio file to YouTube and click the box that generates a transcription. This is completely free — just be sure to mark your file as “private” if you’re just uploading to get the transcription. The singular caveat is that you there is a level of uncertainty on the accuracy side, that you still require you to confirm by checking.

Transcribe is another computer tool that can assist: https://transcribe.wreally.com

As the site suggests: Transcribe offers an audio player that's tightly integrated with a text editor on the same screen. In other words – you hear the text, and you type it. 

The plus side is that there is no more switching back and forth between the audio player and the editor.

Google Docs offers a speech to text option. It appears to work well and a big plus is that it is free. You can use it manually or go to: https://support.google.com/docs/answer/4492226?hl=en

Dragon voice recognition software is also another voice recognition software with potential; again, you have to consider the set up time to ensure the most accurate transcription – and this does not include the interviewees voice recognition.

Other options include:

Transcription services

Large transcription companies can give you cheaper rates if you’re not committed to working with a single transcriptionist. A few popular options include:

Transcribeme:  http://transcribeme.com

Offers transcription delivery within one to seven days. A 10-minute recording with two speakers and a two-to-three day turnaround, for instance, costs about $20. Longer recordings requested with one-day turnaround will be the most expensive, around $4 a minute.

Scribie:  https://scribie.com/

 This service can accommodate almost any project. Its express service, with turnaround in 8-12 hours, costs $3.00 per minute. If you can wait up to a month for your completed transcription, you can pay as low as $.80 per minute.

Rev:  https://www.rev.com/transcription

Rev’s pricing is the most simple: just $1 per recorded minute with 24-hour turnaround. The only extras that cost more are adding timestamps or requesting verbatim transcription, in which you’ll see all of the “mmhmms” and “uhhs” that were recorded.

Voice Base:  https://www.voicebase.com/

This service prices based on turnaround time and accuracy. The most expensive level, which guarantees 99% accuracy and includes speaker identification, costs $2.00 per minute with a five-to-seven-day turnaround. That rate goes up to $2.50 per minute if you need the file in a maximum of four days. Need it in a day? That’ll cost you $3.00 per recorded minute.

So, all in all, the best solution is likely a combination of the above based on your own time constraints, how quickly you will need the information, and whether or not you want to pay to have the services provided by someone else.

Lastly, I do know that a one hour interview will generate 20 -30 pages of typed notes, so taking time to identify the pros and cons of an approach is well worth the time and effort it will take you.

Hi Tim,

I have just read your post.  I am currently in my final year doing an undergraduate degree.  I have had to undertake a wide range of research for my dissertation. I have used a company to transcribe the interviews I have recorded and also to transcribe my dissertation.  I was amazed with the transcription.  It was not perfect but it was better than I expected. I would definitely recommend it.

https://go-transcribe.com/

Hopefully this helps, give it a try.

Hello Tim, 

A friend and I just created a simple web that provides very fast automatic transcription. It has the best transcription tech out there and it is super simple. Right now you can test it for free at:  

Happy Scribe - http://www.happyscribe.co/r/researchgate

I hope it helps!

Marc

http://www.happyscribe.co/r/researchgate

Marc: Thanks for sharing - I will pass it along. 

I used http://www.automaticsync.com/captionsync/ and https://www.tranzify.com/ is very good as well.

I have used Dragon software, the cheap one that is available to download as an app on your ipad. This works fairly well, but so does the Google Docs app freely available as well. I did get a department to purchase a more expensive Dragon package and this actually worked the best of all, however, for the cost of the expensive software I was unsure of the value. I still needed to read and check for accuracy, and I needed to consider whether verbatim transcription was fully needed. 

I am going back to Google Docs version, purely because it produces the text in a word format that I can easily send from drive to drive in a windows series of machines, and it avoids having to keep changing from mac to windows...even though this is becoming very simple.

Has anyone used Happy Scribe for transcriptions?  If so, how did you like it? 

https://transcribe.wreally.com/

Thank you,

Patricia

Thanks everyone for these important information

Transcribe, an online transcription and dictation software is working really well for me now and it makes the process much faster than having to type everything.

https://transcribe.wreally.com/ 

Patricia

Just started using a Microsoft surface to both record and transcribe on. Using a two headphone/ mic jack and working through transcripts quite efficiently. Aids initial analysis and saves a lot of money and time. Have tested out Dragon software.. but also finding that Google Docs works really well with the Surface.

I used ASTE and was very pleased https://web.automaticsync.com/

Spoken ONLINE (www.spokenonline.io) is an automated transcription tool that we launched just recently.

Just upload your audio file and we’ll notify when you can download the ready made text as Word format. You can test the tool with 30 minutes free trial.

I know it can be time consuming but the process is important. As I have said before I am moving to using a headset and microphone and transcribing through Google Docs, therefore ending up with a Word Doc to work on. Unfortunately, listening and transcribing in some form is generally the best way, but a good debate to read is offered in an old paper by Halcomb and Davidson (2006) related to verbatim transcription.

Has anyone tried using Trint for transcribing- looks like it could work well and not cost too much. I'd be interested to know if anyone has tried it? thanks

Hi Oonagh,

I've tried a lot of transcription tools already and I find HappyScribe (https://www.happyscribe.co/) the fastest, cheapest and most accurate so far. It even recognizes over 100 languages. Might be worth a try. :)

Thanks everyone! I will try these new ones out. I used transcribe the last time which was good but fairly expensive.

Bambi

Well, We are bringing ease of use and great UX at

Spext - ( https://www.spext.co )

There are many power features, and more are coming soon. You can even take notes of your audio data. We are getting popular with researchers.

- Ashutosh

https://spext.co

I had a look at oTranscribe and it seems great, but I have a problem with browser-based software: does that mean that my files will be uploaded to someone else's server? That raises some privacy issues...

Any transcription service will have that issue. If that is of concern for you (my IRB allowed with privacy notice from the service I used) then you will have to transcribe yourself and keep all audio/text on your machine.

Best in all you do!

Thanks for both your answers. I think my hands are tied in that sense because of the personal data policy of my ethics committee...

If you must transcribe yourself, get a foot pedal - it really helps slow down speech and repeat when needed.

Thanks for all these suggestions. My humble opinion is, if you have a lot of interviews, you need to pay for a good service (several recommended in this thread). If you have time, and feel it important to give your own clarity to transcriptions, do it yourself and use speech to text in Word. Not sure if you can do this on PC in which case use the Google Doc version of this. Use VLC or similar to steadily work through each audio file, repeating back into the dictation mic what the participant says. It is by far the most accurate method. The time you waste trying other methods you may as well just get on with it! Tip: remember to timecode periodically in the transcripts - it's super useful.

FYI I have also just experimented with the Soundflower app method using Audacity and Word on Mac but it mostly returns a jumble of meaningless text.

NB - keystrokes for turning on and off the speech-to-text in Mac are 'fnfn' for on, then 'fn' (or 'fnfn') for off. This really speeds up the process. Keep the two softwares side by side and just click between them. It's much faster than typing by hand and keeping software workspaces full screen. (Don't mean to patronise but jus giving my experience.)

Easytranscript, quite like f4 but for free:

http://www.e-werkzeug.eu/index.php/de/produkte/easytranscript

I think Dragon Speech Recognition is the best transcription software.

I tried Dragon for transcribing my two person interviews and it didn't really work at all. It did a decent job with my voice but produced absolute gibberish for the other speaker. Given my interviews were 60 - 90 minutes long, the transcript was often 20 - 30 pages long. It took me hours to produce a decent transcript. I switched to Rev.com and couldn't be happier! While it is certainly more expensive - $1/minute - the quality and turnaround time are more than worth it to me.

https://www.rev.com/

I wanted to update my earlier experiences using Microsoft Word and VLC, and the speech to text method. I have discovered through transcribing around 10 interviews (so far!) that several voice activated punctuation and grammar commands help a lot for transcript layout. I list what I have found out below. These are activated by speaking them into the speech to text of Word.

• New line (a new line starts)
• Comma
• Full stop
• quote ... quote
• Open bracket/close bracket

Ill add to this list as I discover new commands.

If you plan on recording and transcribing interviews or meetings past your university experience, I have found that Nuance's Martel Multi-Voice Recognition Recorder automatic Voice to Text System is a great choice. It is called the Ghostwriter. Martel boosts that it is the world's first of its kind multi-voice system and it comes with a heft price tag but it gets the job done effectively and efficient. The system comes with a high-quality Philips recorder with charging and uploading docking station, two individual microphones (8 can be added) with 3D Mic technology, and uses Dragon software. The finished text document can be edited after automatic transcription easily using both voice commands and manual editing, then saved in Microsoft Word documents. Martel also provides excellent support (my experience).

http://www.martelelectronics.com/worlds-first-multiple-voice-recognition-recorder-conferences-meetings-automatic-voice-to-text-system/

Some corrections made to my previous post: If you plan on recording and transcribing interviews or meetings past your university experience, I have found that Nuance's Martel Multi-Voice Recognition Recorder Automatic Voice to Text System is a great choice. It is called the Ghostwriter 3. Martel boosts that it is the world's only multi-voice system and it comes with a hefty price tag but it gets the job done effectively and efficient. The system comes with a high-quality Philips recorder with charging and uploading docking station, two individual microphones (8 can be added) with 3D Mic technology, recognizes various languages, and uses Dragon software. The finished text document can be edited after automatic transcription easily using both voice commands and manual editing, then saved in Microsoft Word documents. Martel also provides excellent support (my experience).

http://www.martelelectronics.com/worlds-first-multiple-voice-recognition-recorder-conferences-meetings-automatic-voice-to-text-system/

📷

The balance between cost and time is important. There are many systems that present extremely good functionality. I come from the side of using freely available software that many of use have in our devices, and exploring if they have some utility. In a project with a colleague (Angelina Chadwick) we found that the simple Google Docs through a Microsoft Surface had some utility. We do not advocate that it is by any means the best system, however, it is free and simple to use. We have also worked with Apples versions, each equally simple and with time having utility. Manual editing is the main problem once you have got used to the system.

Man I tried most of these that were posted but this by far was the best one. Plus it's FREE!

https://support.google.com/docs/answer/4492226?hl=en

Follow the directions for google docs and you're in business.

yes agree with previous quote totally. i think, in the words of a colleague, the transcript process is like a pre-analysis phase.

Rocky Dwyer

Hi,

I would like to know about the ethical aspects of using the software for transcribing. Even if we use Youtube in private mode, to upload the transcription, the audio file needs to be changed into a video file and to do that we may have to use any online converter.

So I just want to know will it anyway harm the ethical aspect in the research.

Thank you

I would suggest that ethics are directly impacted by aspects such as:

1. Is the interviewee or interviewer identifiable by the interview audio?

2. How sensitive/personal is the information being discussed?

3. For what purpose is the interview - that is, is it for security, financial or other similarly 'sensitive' institution/policy development etc?

4. Who's account are you logged in with to the cloud software?

5. Does the cloud software retain the data? If so, how long for? Can you clear your data history?

Choosing desktop (local) software is probably a better option if you feel ethical issues outweigh ease of use of a cloud service.

Nvivo just announced its new, coming soon transcription program. If you are interested in finding out more about it, follow the link below:

https://www.qsrinternational.com/nvivo/nvivo-transcription?utm_source=Pardot&utm_medium=email&utm_campaign=GBL%20-%2018%20-%20Jul%20-%20NVivo%20%20Transcription%20Pre%20Release%20Email%201&utm_content=Landing%20Page%20Link%20Users

The forthcoming NVivo transcript program promises 90 percent accuracy, which is about one error per sentence.

An error per sentence is not very impressive. A person who makes an error per sentence in their native language would be considered to have a speech impairment.

Most of the new transcription services, which leverage technologies based on AI, get better the longer a file sample is and typically start with better than 95% accuracy for < 1 minute long audio files, hit 99% for ten minutes and start adding decimal points after that.

Personally, I use Raven's Eye, mentioned above. It is cheap, accurate, and what I came across first, but there are other solutions that offer similar accuracy. Most of them leverage the SaaS transcription back-ends provided by the major players in tech: Google, Amazon, IBM and others.

Atlas ti is very user friendly and I have been using it and found very easy to generate coding and network building.

In answer to concerns about capturing non-verbal cues and the importance of transcribing as the first analysis: I used automated speech-to-text to pump out the first version of a text, then I went through manually and edited it. Quicker than doing it all manually but I did get the advantage of a first analysis. I recommend it.

Mary, what program do you use for the automated speech-to-text?

We got lot of people from this forum to try out https://spext.co for the transcription need. I am glad to share that we have been ranked #3 in top transcription products ahead of other 20 in this category. Here is the link to the benchmark - https://blog.timbunce.org/2018/05/15/a-comparison-of-automatic-speech-recognition-asr-systems/

My methodology (phenomenography) and other interpretivist approaches value the first level analysis of transcription and will even document its process, or commentary on it (e.g. Dortins, 2002). Some place importance on non verbal cues, others on wider context. When you transcribe manually (using speech to text to speed up writing) you do get an understanding of each interview, its general 'feel'. I don't think automated processes can help with that at all.

You should try Scribie (scribie.com). It is very accurate in Spanish and English, and they offer different plans.

I had a relatively positive experience using NVivo the first time I used it. I liked how it automatically breaks up the text based on the speaker, and adds the times for you. Well, it attempts to do that, but doesn't always get it right. There are lots of mistakes that you have to go back and fix, but I don't think any product on the market can help avoid that. Another upside is that the product has a simple interface. One downside to NVivo is that it is pricey: $30/hr if you only use a few hours, with the price decreasing to $20/hour if you buy hours in bulk. There currently is no subscription option.

However, I later ran into problems with Nvivo. I think the issue might owe in part to my file being much longer the second time. The resource froze up frequently, and was very buggy overall.

To answer Jane: I used temi.com

I had to come back and add this. I just used Happy Scribe for automated translated but I also used the in-tool editor to edit the text. The editor is a huge help! Previously I have taken the transcripts out and then edited manually in Word and using Windows' own player for the audio. Happy Scribe's editor lets you simply put the cursor on a piece of text and it takes you right to that spot in the audio. It's so much easier to go directly to the spot where you need to hear the audio again. It's made my process much faster. It does cost something but my first experience has been very good and efficient.

For those of us who like to get a good feel of the data as we transcribe by hand, Im now using a more streamlined method to that which I outlined in earlier comments. Im using Evernote to do the first pass of hand transcription, with the audio track embedded in the note. This makes it much easier to pause and start the audio while typing. You can still the inbuilt use voice to text and I wish I had thought of this earlier!

Mary Nurminen how long were your interviews transcribed by Happy scribe? My interviews are approximately 90 minutes and I wondered what on line transcription service manages interviews of this length?

thanks

Maureen

Maureen MacGinley , some of my interviews were 90 minutes and there was no problem with that, Happy Scribe managed them just fine. I don't know, but I suspect that others have no problem with that kind of length either.

thank you Mary. This is very helpful.

Hi everyone, and thanks for the very helpful threat. I am about to leave for fieldwork and I have some questions:

- Does it make sense to buy a recorder with "transcription technology" included, if softwares are usually separate and work on audio files once they've been transferred on a computer?

- What audio file is used with the softwares you mentioned? Is it .mp3 or other formats?

- Is there a way in which audio files can be encrypted already on the recorder?

Thank you very much for any tip on this!

I don't think it makes sense to pay for technology that you may not use. Instead, you should the money and spend it on software that clearly meets your needs.

In terms of what file format is used, most software will read a wide variety of formats, including both .mp3 and .wav.

Although I have no experience with recording encrypted files, I have never seen this offered on any of the recorders I have examined.

Overall, if you are simply recording one-to-one interviews, I would recommend avoiding spending money on features you don't need. In particular, most recorders are designed to maximize their ability to record voices, and paying more usually buys you the ability to record music at high quality levels. I have used quite a number of digital recorders over the years, and I have found some very good ones at about $60 USD.

Thank you David L Morgan ! I also did not intend to pay more than that price for a recorder, still there are some with transcription included (Olympus WS-852 for instance) and I was not sure of what was the advantage. If you have any suggestion in particular, for both recorder and software, it’s most welcome. The encryption issue was to protect the privacy of interviewees, maybe there is an easier way to lock the content on recorders. Thank you again for your expertise!

I have used Olympus recorders that are earlier versions of the WS-852 with good results. According to Amazon, it costs $50.99 which is reasonable. If you are very price sensitive, then I would recommend one of the non-stereo versions of the same recorder. The main use I have stereo is when there are two or more participants, because the stereo makes it easier to separate the participants according to their locations around the recorder.

I have been using "transcribe" https://transcribe.wreally.com/ in the last few weeks, which requires subscription. It was suggested in one of the comments in above, however, I found it somehow helpful but not significantly, although it says it has 90% accuracy.

It costs $20 for subscription and $6 each time the person buys minutes (the $6 gives only 60 minutes and the minutes go quite fast). Then, a friend recommended Otter http://otter.ai which offers 500 free minutes and is significantly helpful.

Remember, 90% accuracy implies an error very 10 words, or approximately 25 errors on a double-spaced page.

Whatever software you use to transcribe requires you to check the text again especially if recording someone with poorer english pronunciation which the software cant recognize..

mp3 is a compressed file format - it's fine for distributing. Audio recorders such as the ZoomH6 which I use for recording podcasts recording in high res audio. My half hour shows are typically 500MB plus in raw format. I then edit in Garageband on Mac, and export as a much smaller .mp3 files, typically 50MB...

I agree with Jasmin. I have tried to trascript for a free trial, and with accents (may be not clear English), it was terribly reproduced.

Thanks everyone. This is very helpful.

Experimenting with transcription for Hebrew multi-participants interview - very tough conditions. Both transcribe.wreally.com and listenbycode.com provide very similar results. The results are "better than nothing" - I think about 50% accuracy. I'm attaching both a 1 minutes ample and the corrected transcription results - blue are my additions, red are my deletions, so you can get an idea.

The other observation is that no matter how accurate the transcription is, we read differently from how we speak - so the recording of an interview will need to be edited for it to make sense to a reading audience. This might take some time...

One way I transcribed a multi-participant, multi-accent group was to respeak the whole thing into Google Docs' automatic transcription system (does it while you talk). I don't know if that gave better results that if I'd entered a recording into another automatic transcription system, but I guessed that the multiplicity of accents would have made the results of the latter pretty bad.

Again, softwares are good for help but there is always a need to look into the text and revise it in cases of misspelling.

Kvale (1996, mainly) thinks of transcription as a clear stage of analysis: the data itself changes from verbal to written. I very much believe that doing the transcripts by hand, either using V-to-T by repeating what the interviewee says (as mary above says), or simply typing as you go along in the recording, provides a deeper sense of what is there. No matter what your chosen analysis methodology is, you need to have a deep understanding of both individual transcripts and the data as a whole. Unless you have literally 100s of interviews, I would strongly recommend doing it yourself, or in your team of researchers. It is an aspect of the art of hearing the data (Rubin & Rubin, 2012, excellent book).

Think of avoiding transcription entirely, but analyzing the audio data: Article Structuring Audio Data with a "C-TOC ": An Example for Analy...

I see that it's common to upload the audiofiles to external websites and was wondering if dataprotection/privacy is not an issue in this instance?

I found Otter to be better. It uses AI technology and is much cheaper. You can check it out here: https://otter.ai/referrals/2VMC7850