Originally published by me at The CogDog Show » Syndicate This (see it there)

Sometimes I get lucky and two ideas on different paths cross paths right in front of me. It’s a matter of just reaching out and grabbing them.

Brian had an idea for our You Show Audio unit to add (or maybe replace?) the instructions for the activities with an audio explanation. This is the kind of approach I like to use a medium or technology to explain the medium or technology.

I wondered somewhat about the downside of not making the content accessible, which is not a direct concern, but one of those things that one should keep in mind.

At the same time, in an email group, my colleague from Maricopa, Alisa Cooper, shared a link to an interesting tool- a speech recognition add-on for Google Docs.

My stream crossing idea was to record the audio at the same time running the add-on to see how well it did with transcription. I’ve had the usual mixed bag experiences of garbled YouTube annotations and yelling corrections at Siri.

The tool adds itself to an Add-ons menu in GDocs. I almost gave up, because it never showed me the promised “start” link, but it showed up when I saved the doc, closed, and re-opened it (once again, the Old Reboot Trick works).

I set up my Samson Meteor mic, turned on Audacity to record, then activated the Speech Recognition add-on. I put all of this behind the web page for the activity so I could focus on whatever unplanned commentary I would make (hey I am just testing this out).

It was interesting to see what it came up with:

View of the speech to text right after stopping the recording.
View of the speech to text right after stopping the recording.

There is no response to end a sentence or change paragraphs, so you get the blob of text. It got a good amount and messed up maybe 20%? 25%? Even at that, I sure prefer that than trying to transcribe from the audio (if you have ever seen me type, it may have been an instigating reason for my marriage ending).

Here is the blob that came out before editing it:

okay I’m going to try explaining the first assignment for the new show the unit for I’m recording this but I am also using a new add on to Google Docs has actually transcribing this to text that is not essential force activity sounds of your work the idea is to capture two different kind of sounds that you can use inside of your project next week so this could be the place you work or could be just special place at home but the idea is to think about the environment of the place so I do a lot of my work here Cincinnati are you in my apartment what kind of empty right now usually I have something on the TV I would it on CNN large can background music it’s pretty empty but listen very closely there’s a small low taking excess old fashioned analog clock on the floor and my fridgerator ambient sound so the suggestion that something MySpace which what I would do for my Simon is just recorded it without me talking it’s just to hear the background sound set environmental sound your sound like a perhaps a more noisy office in my feet in a library might be inside of a laboratory somewhere the idea of the same it is to get something that you can use it later today to perhaps a background as you talk about your work but it sounds pretty sure I’m Beyonce or sense of the environment second summer assignment something completely different phones idea I’ve never done this life it’s interesting the promises to record one sound one single sound get something with an inhaler and we’re just recorded just a little bit sound this week it won’t mean much hope for the child’s health Decatur many times space it out make it go up make it go down make a change pacemaker be like a river runs layered on top of its off to make that one sound beyond most like corpus and it’s just an experiment to see how much you can pimples with a single sound like a single note longest time trying to learn how to play harmonica so I’m just going to play a single product keratin out and see what I can do with that and perhaps edit favorite tracks put it in the sound pull up reps upload it to my blog and just keep it for next week so we’re just grabbing raw material this week that’s all I’m doing and I just want to have this is a record

And you can compare it to my post edits in the doc (which I guessed more at than trying to match perfectly to the audio).

In Audacity where I recorded it, I trimmed the in and out points. To demonstrate some techniques, I decided to put some music in as a background, using “Perspectives”, a creative commons licensed track from the most useful Incompetch Royalty Free Music Collection — by the way this site does a stellar job of providing attribution text:

“Perspectives” Kevin MacLeod (incompetech.com) Licensed under Creative Commons: By Attribution 3.0 http://creativecommons.org/licenses/by/3.0/

In Audacity I use one of my favorite bits, the envelope tool to fine tune the audio to duck it under my un-dulcet mumbles. My voice is the top track (I again was too close to that mic), and the music in the lower track.

audacity jon edit

In the You Show page I am able to embed the audio, add a download link, and a link to the transcript I did in Google Docs


DownloadGoogle Docs transcript

It’s not perfect, but was worth experimenting with.

While looking for more information, I stumbled into Online Dictation, a web based speech to text generator. I was fascinated to watch how the words morphed and change as I spoke. It seemed to do a little better, and I was able to have it end a sentence by trying “dot” (or was it when I said “period”)? Here is a little screen cap of my test.

Actually this would be pretty neat for doing some mashup type videos.

Stuff just happens all the time, crossing streams and everything. On the net, its more than fine to cross the streams

The CogDog Show » Syndicate This is a portfolio site built for my time as an Open Learning Scholar at Thompson Rivers University.

No comments yet.

Leave a Comment

All fields are required. Your email address will not be published.