Annotape: Audio Data Collection and Analysis
Allows a user to record audio interviews and analyze qualitative data.
AnnoTape is a solution for recording, analysing & transcribing audio data for qualitative research, marketing, journalism and broadcast, and sound archiving.
AnnoTape turns your computer into a ‘virtual’ tape recorder:
One hour of recording normally takes between six to eight hours of professional transcription, involving time and expense. Instead, cut out transcription all together - AnnoTape lets you record, review, analyse and index the sound files themselves.
The design team for AnnoTape brought together a software engineer with an anthropologist specialising in qualitative research techniques. Amongst its present uses, AnnoTape has been tested in ethnographic fieldwork, in market research and focus group management, in qualitative research in the health sciences, and other research activities. It can be used in any environment where recording, storage and manipulation of large amounts of text and audio data are necessary.
How does AnnoTape work?
AnnoTape name describes how the programme is conceived: in essence, AnnoTape works with ‘annotated tape’. Think of it this way: imagine recording an ordinary cassette tape of you interviewing someone for, let’s say, an hour. Imagine, further, that you could stick tiny pieces of yellow stickie paper along the tape marking moments where a particularly interesting discussion begins: let’s say your interviewee has just started, at moment 10:11:40, to discuss the heart of the subject you are interested in. Months later, when you have probably collected tens or hundreds more such cassette tapes of interview, you know you are going to want to listen to this moment again - but how will you ever find it?
In essence, AnnoTape allows you to turn your computer - preferably a laptop for mobility, but desktops work too - into a recording unit. You can record hundreds of different interviews or other kinds of sound file straight onto your hard disk (or, if you are of the old school, use a conventional cassette recorder and then upload the recordings to the hard disk later). All of these sound files are then instantly available for playback at any moment in the future.
In the second step, you listen through each sound file after it has been recorded, and you ‘annotate’ it - that is, you stick virtual pieces of yellow stickie paper to particularly striking, intriguing or useful moments in the sound file to which you know you will want to return later. The ‘stickie’ is an index file to which you can give a name or sequence of names - these may be unique to that one place in that single file, or may be an index term which you use across all the files in your archive. Either way, these indices are fully searchable later. You can also attach free text ‘index notes’ to this moment in the ‘tape’ also.
In this way, you build up a database which consists of all the sound files (and text files, which can be AnnoTated in a similar way), together with all the indices which point to particular moments.
At some point in the future, you may be writing a research paper or report, preparing or giving a lecture with audio-visual aids, or simply wanting to check your recall of some recorded events. At this point, you can search for and instantly retrieve any moment that you previously indexed. Say, for example, you know that somewhere in your one hundred hours of recorded interviews, you know you previously indexed a moment where one informant (you can’t now remember who…) referred to the problems of teenage tobacco addiction. Using the find command, you pull up a list of all the index terms you have attached, and you see that one of them is ‘Tobacco:Addiction:Teenagers’. You search on that term, and immediately, AnnoTape finds that you applied that term just once, to the file with the interview with ‘Patrick’, beginning 30:35:20 into the interview. Press return, and the recording starts to replay at exactly that point.
Finally, AnnoTape was developed precisely so that laborious transcription of sound files would be a thing of the past. Since transcription takes so long, you want to avoid it if at all possible. By allowing you to do all of your analysis and review working with the original soundfiles, AnnoTape achieves this. However, at the end, when you have culled perhaps twenty or a hundred key moments from the sound files that you think sum everything up for your project, you may have need to transcribe short bursts - five minutes here, two sentences there, etc. AnnoTape’s in-built transcription editor allows you to pace backwards and forwards through the tape, stopping and starting, so that you can transcribe what you need without ever leaving the programme.
FOR MORE: As always, the best way to learn is by doing - download a Demo copy of AnnoTape from the website now and investigate the possibilities. Alternatively, take a look at the essay on this Website called ‘Using AnnoTape for anthropological research in the Eastern DR Congo’ which describes the use of AnnoTape in a long-term research project environment.
Does AnnoTape automatically transcribe any voice recordings (does it, in other words, perfect speech recognition technology?)?
A resounding NO! In short, AnnoTape is not a universal transcription engine - something which still remains impossible. Instead, it makes universal transcription unnecessary.
Voice transcription is the Holy Grail of computer-aided audio research. At present, a number of packages exist commercially which can, with time, learn to transcribe the speech of a single user - i.e. you. Equally, some packages exist which can discriminate between a small, strictly limited number of words - let’s say the numbers one through ten - for pretty much any voice and accent.
However - and this is a HUGE however - there is a big difference between these types of application and the ideal of a computer which can transcribe any voice, with any accent and any pitch, under possibly poor recording conditions and speaking on any subject under the sun. Though we regularly see such things enacted on shows like Star Trek and so on, this ‘unlimited’ form of speech recognition is still well beyond the bounds of possibility for now - estimates suggest that it may be twenty years before this ‘Holy Grail’ is achieved.
In the worlds of interviewing, qualitative research, ethnography and so on, full text transcriptions of sound files are used so that review, analysis and indexing can be carried out. But manual transcription - the only available option at this stage because of the limitations discussed above - takes between six and eight hours per hour of recorded tape. For most people this is simply unmanageable.
AnnoTape was invented precisely because of this problem. Rather than trying to solve the almost impossible task of universal voice recognition, we developed AnnoTape so that all of the review, analysis and indexing could be done with the original sound files, making initial transcription unnecessary.
So, in short, AnnoTape is not a universal transcription engine - something which still remains impossible. Instead, it makes universal transcription unnecessary.
What do I need in order to run AnnoTape?
Windows 98 or NT 4.0 or later, with a minimum of 32 Mb of internal memory.
In addition, AnnoTape requires the following:
Basically, it requires a Power-Mac based system running system 7.6 or later, with a minimum of 32 Mb of internal memory. A laptop (PowerBook G3, for example) is particularly convenient as it can then be used as the portable recording device.
In addition, AnnoTape requires the following: