Hello,
I would like to ask if it’s possible to import a tab delimited text file aligned to the audio file. I have texts and audio files, but the text files are not aligned to the audio files. Moreover, my audio file are dyadic conversations, so I have to deal with two different speakers.
Are there any solutions?
Thank you in advance!
E.
Hello,
The options you have depend on the language spoken in the audio files and on how the texts are organised.
In principle it is possible to import tab delimited text files in ELAN. If the file does not contain time stamps, the user can specify an estimated default duration per sentence (i.e. per row in the tab delimited text file). After also linking the audio file, the annotations (the segments) can manually be aligned with the audio: a time consuming task. In this scenario it would be best if the texts are already stored separately per speaker.
ELAN also offers interfacing with several web services that might be of interest. These services are listed in the Recognizers tab in the ELAN user interface. The most promising is probably the BAS WebMAUS service. With the audio file loaded in your ELAN window, you could browse to your text file as the Orthographic text file parameter. I guess in this situation it would be best if the text of both speakers are in one file (not sure about this) and I’m also not sure if tab delimited text is supported.
There are other recognizers, like the Speaker diarization pipeline, that try to automatically segment the audio into segments per speaker, but without producing the text. You would then have to transfer the texts you have to those segments, manually again (as far as I can see).
Finally, I mentioned the BAS WebMAUS service. The BAS services have more options online, in the web browser (you can follow this link). There might even be an ASR (Automatic Speech Recognition) option available for your language.
-Han
Even i am facing the same issue , The CSV tab delimited text file is not aligned to WAV audio file .
The transcriptions by ASR are not aligned to audio files .
However the solution provided doesn’t seem to help me as i am unable to find the recognizers tab in the ELAN user interface . I need a solution to fix the ELAN issue .
Need help on this .
Hello,
The Recognizer tab is visible in the Annotation Mode (Options->Annotation Mode), in case you’re usually working in a different mode.
Maybe you would like to sent some screenshots and, if possible, example files to me (han.sloetjes AT mpi.nl), so that we can try to find a solution for the different problems?
-Han