Re: [CODE4LIB] audio transcription software
Eric, I tried Docsoft:AV (http://www.docsoft.com/Products/AV/), a server-based solution, about a year ago to see whether we could use it to automatically transcribe and timestamp our oral history recordings. It might work nicely if you had multiple recordings with the same speakers where it would be feasible to train the software by setting up speaker profiles for the individual speaker's voice. The software can output the results in a variety of formats and it handles audio and video recordings. However, we only had one recording per interviewee (intermixed with the interviewer) and thus would have had to spend way more time and money on training the software (and cleaning up the results, which were hardly comprehensible) than if we had an actual person listen to the recordings and transcribe them. To be fair to Docsoft, some speakers had strong accents and the audio quality was not ideal, but that's what we needed it for. So, it did not seem to be a feasible solution for this particular problem and we stuck with a wetware-based approach. Markus Markus Wust Digital Collections and Preservation Librarian Digital Scholarship and Publishing Center North Carolina State University Libraries
Re: [CODE4LIB] audio transcription software
I've saw a reference to some software called IBM ViaScribe when reading about a project that converts lectures to text (http://www.liberatedlearning.com/technology/index.shtml) a while back. Kristina CODE4LIB automatic digest system wrote: Subject: CODE4LIB Digest - 11 May 2010 to 12 May 2010 (#2010-115) From: CODE4LIB automatic digest system lists...@listserv.nd.edu Date: Wed, 12 May 2010 23:00:16 -0400 To: CODE4LIB@LISTSERV.ND.EDU To: CODE4LIB@LISTSERV.ND.EDU There are 12 messages totalling 394 lines in this issue. Topics of the day: 1. audio transcription software (12) -- Kristina Long Programmer reSearcher Software Suite -- researcher.sfu.ca Simon Fraser University Library kl...@sfu.ca
Re: [CODE4LIB] audio transcription software
On May 12, 2010, at 2:24 PM, Jonathan Rochkind wrote: Do you mean software to aid a human transcriber, or do you mean software that can actually use voice recognition to turn audio to text all automated? I am interested in the later -- software that converts audio files in to text files. -- Eric Lease Morgan
Re: [CODE4LIB] audio transcription software
I tried Dragon Naturally Speaking a couple of years ago. (After breaking a wrist in a cycling accident, I figured it might be easier than one-hand typing, which wasn't true in the case of typing programming code with lots of curly brackets, indentation, etc.) Speech-to-text software works best after a training session, in which the software asks the speaker to read a known text, to calibrate the software. I'm not sure how it might work to calibrate for voices on recordings, but it may be that the software can learn during a proof-reading process. Your success for oral history recordings may depend on the uniqueness of each speakers voice, and the length of each recording. (Lots of short recordings of many different speakers would tend to be harder.) Keith On Wed, May 12, 2010 at 2:18 PM, Eric Lease Morgan emor...@nd.edu wrote: Does anybody here use or know of any audio transcription software? We have a growing number of projects here at Notre Dame that include oral histories. How can these digital files be converted into plain text? Audio transcription software may be the answer? -- Eric Lease Morgan University of Notre Dame
Re: [CODE4LIB] audio transcription software
Google Voice transcribes voicemails, but I don't think there is any api to use it outside of their system. I also haven't used it much so I don't know how accurate it is. Joel -Original Message- From: Code for Libraries [mailto:code4...@listserv.nd.edu] On Behalf Of Eric Lease Morgan Sent: Wednesday, May 12, 2010 2:30 PM To: CODE4LIB@LISTSERV.ND.EDU Subject: Re: [CODE4LIB] audio transcription software On May 12, 2010, at 2:24 PM, Jonathan Rochkind wrote: Do you mean software to aid a human transcriber, or do you mean software that can actually use voice recognition to turn audio to text all automated? I am interested in the later -- software that converts audio files in to text files. -- Eric Lease Morgan
Re: [CODE4LIB] audio transcription software
Not software, exactly, but this seems like an ideal thing to set up in Mechanical Turk. This guy did it with an audio interview: http://waxy.org/2008/09/audio_transcription_with_mechanical_turk/ -Sean On 5/12/10 2:18 PM, Eric Lease Morgan emor...@nd.edu wrote: Does anybody here use or know of any audio transcription software? We have a growing number of projects here at Notre Dame that include oral histories. How can these digital files be converted into plain text? Audio transcription software may be the answer?
Re: [CODE4LIB] audio transcription software
A great quality service is http://on-sitemedia.com/. I have no idea what she charges. Google for open source speech recognition On Wed, May 12, 2010 at 1:05 PM, Sean Hannan shan...@jhu.edu wrote: Not software, exactly, but this seems like an ideal thing to set up in Mechanical Turk. This guy did it with an audio interview: http://waxy.org/2008/09/audio_transcription_with_mechanical_turk/ -Sean On 5/12/10 2:18 PM, Eric Lease Morgan emor...@nd.edu wrote: Does anybody here use or know of any audio transcription software? We have a growing number of projects here at Notre Dame that include oral histories. How can these digital files be converted into plain text? Audio transcription software may be the answer? -- --- www.maf.org/rhoads www.ontherhoads.org
Re: [CODE4LIB] audio transcription software
youtube actually does this automatically (but only in english). but i'm not going to lie, it sucks. however maybe that's just my voice since i'm always stuffed up. On Wed, May 12, 2010 at 3:24 PM, Brad Rhoads bdr...@gmail.com wrote: A great quality service is http://on-sitemedia.com/. I have no idea what she charges. Google for open source speech recognition On Wed, May 12, 2010 at 1:05 PM, Sean Hannan shan...@jhu.edu wrote: Not software, exactly, but this seems like an ideal thing to set up in Mechanical Turk. This guy did it with an audio interview: http://waxy.org/2008/09/audio_transcription_with_mechanical_turk/ -Sean On 5/12/10 2:18 PM, Eric Lease Morgan emor...@nd.edu wrote: Does anybody here use or know of any audio transcription software? We have a growing number of projects here at Notre Dame that include oral histories. How can these digital files be converted into plain text? Audio transcription software may be the answer? -- --- www.maf.org/rhoads www.ontherhoads.org
Re: [CODE4LIB] audio transcription software
depending on your budget there are quite a few services available to do it for you. some include time-code information depending on what interfaces you want to build. came across this list in someone's delicious feed on here: http://www.uiaccess.com/transcripts/transcript_services.html eby On Wed, May 12, 2010 at 2:18 PM, Eric Lease Morgan emor...@nd.edu wrote: Does anybody here use or know of any audio transcription software? We have a growing number of projects here at Notre Dame that include oral histories. How can these digital files be converted into plain text? Audio transcription software may be the answer? -- Eric Lease Morgan University of Notre Dame
Re: [CODE4LIB] audio transcription software
On Wed, May 12, 2010 at 11:46 AM, Keith Jenkins k...@cornell.edu wrote: I tried Dragon Naturally Speaking a couple of years ago. (After breaking a wrist in a cycling accident, I figured it might be easier than one-hand typing, which wasn't true in the case of typing programming code with lots of curly brackets, indentation, etc.) Funny you should mention that. About a week ago, I stuck my hand between two with large grouchy dogs rendering my hand and wrist useless. So I downloaded the free voice recognition software that can be downloaded for Windows XP and is built into Windows 7 since my hand and wrist still hurt like heck. I spent about an hour training it, and my conclusion is that it sucks but it is better than typing one handed, so I am still using it (including to type this message). Download SpeechSDK51.exe from the Microsoft Speech SDK 5.1 if you're interested. It's totally free and worth every penny. kyle
Re: [CODE4LIB] audio transcription software
I'm embarking on exactly the same thing, and for films as well. Depending on the fidelity of your recording you will get different results using Premier Pro in the Adobe Creative Suite 4/5. For modern (2009 recordings) high-fidelity audio we only got 80% accuracy. Dragon still does not work in Keith's scenario. Dragon works ok if you are echoing recorded speech, that is if you repeat the recording using your voice. The production rate is a little bit better than that of a novice typer, so if you have an army of students you might make some headway. I vowed never to attempt this again after echoing a 60 minute film. In the end we are outsourcing the bulk of our transcriptioning. John -Original Message- From: Keith Jenkins [mailto:k...@cornell.edu] Sent: Wednesday, May 12, 2010 2:47 PM To: CODE4LIB@LISTSERV.ND.EDU Subject: Re: [CODE4LIB] audio transcription software I tried Dragon Naturally Speaking a couple of years ago. (After breaking a wrist in a cycling accident, I figured it might be easier than one-hand typing, which wasn't true in the case of typing programming code with lots of curly brackets, indentation, etc.) Speech-to-text software works best after a training session, in which the software asks the speaker to read a known text, to calibrate the software. I'm not sure how it might work to calibrate for voices on recordings, but it may be that the software can learn during a proof-reading process. Your success for oral history recordings may depend on the uniqueness of each speakers voice, and the length of each recording. (Lots of short recordings of many different speakers would tend to be harder.) Keith On Wed, May 12, 2010 at 2:18 PM, Eric Lease Morgan emor...@nd.edu wrote: Does anybody here use or know of any audio transcription software? We have a growing number of projects here at Notre Dame that include oral histories. How can these digital files be converted into plain text? Audio transcription software may be the answer? -- Eric Lease Morgan University of Notre Dame
Re: [CODE4LIB] audio transcription software
It's not great. In my experience, it gets some words right. Mike -Original Message- From: Code for Libraries [mailto:code4...@listserv.nd.edu] On Behalf Of Joel Marchesoni Sent: Wednesday, May 12, 2010 11:49 AM To: CODE4LIB@LISTSERV.ND.EDU Subject: Re: [CODE4LIB] audio transcription software Google Voice transcribes voicemails, but I don't think there is any api to use it outside of their system. I also haven't used it much so I don't know how accurate it is. Joel -Original Message- From: Code for Libraries [mailto:code4...@listserv.nd.edu] On Behalf Of Eric Lease Morgan Sent: Wednesday, May 12, 2010 2:30 PM To: CODE4LIB@LISTSERV.ND.EDU Subject: Re: [CODE4LIB] audio transcription software On May 12, 2010, at 2:24 PM, Jonathan Rochkind wrote: Do you mean software to aid a human transcriber, or do you mean software that can actually use voice recognition to turn audio to text all automated? I am interested in the later -- software that converts audio files in to text files. -- Eric Lease Morgan
Re: [CODE4LIB] audio transcription software
I've had modest success with MacSpeech Dictate. It uses the same engine as Dragon Naturally Speaking, but has none of the canned read-back training that Dragon requires. I've been testing it with good quality recordings of our local veterans.(already in digital format on CD) The process I've had most success with is to let MacSpeech Dictate take a run of about 10 minutes, then go back and correct the resulting transcript. Next I re-assign the transcript as the learning text that MacSpeech Dictate uses as it re-listens to the same 10 minute segment of the veteran recording. After this initial run, the accuraccy climbs from the starting 50-60%, up to about 80-85% ... still not perfect, but *good enough* to generate text for indexing and keyword searching. If the transcript is going to be exposed/public facing rather than just machine use, then my next step would be to assign the audio file and MacSpeech Dictate transcript to a volunteer/student for final edits... but at least the starting transcript for human editing would be well along the way.
