So, recently I decided it would be neat to have the computer read me back a talk I had written and given. I suppose I could just record it by reading it aloud into a microphone, but that would get tedious and who likes to hear their own voice? Why can't a Siri-like voice could read it for me? Hmm
I did this on a Ubuntu Linux desktop computer, but I suppose you might be able to do similar things on another system .. but I leave that to you. This tutorial is mostly to document for myself what steps I took to accomplish this task - so that I might be able to repeat my success sometime in the future. It presumes an intermediate level of computer knowledge.
Basically, I wanted the computer to speak the written text to my speakers, and then capture that output directly back into a audio recording program and output the result into an mp3 or ogg audio file. Can we do it? Sure!
Step one: I wrote out the text. In this case, I had done this earlier in preparing to give my talk, one I recently gave as church talk and had concurrently posted on my talk archive blog (in text format). But how to convert those written words back to a spoken form, somewhat automatically?
Basically, I wanted the computer to speak the written text to my speakers, and then capture that output directly back into a audio recording program and output the result into an mp3 or ogg audio file. Can we do it? Sure!
Step one: I wrote out the text. In this case, I had done this earlier in preparing to give my talk, one I recently gave as church talk and had concurrently posted on my talk archive blog (in text format). But how to convert those written words back to a spoken form, somewhat automatically?
- I use free open source Libreoffice text processor, but any text editor would do.
- For my blog I use Weebly .. but that is for another tutorial ..
Step two: I cut and paste the text into the free (and magic) online website form at ttsreader.com advertized as: "A [text to speech] web app that works great both on Chrome and Safari. No need to download or install. No annoying login either. Simply paste the text and click "play". Sweet! Cut, paste, click play, listen .. so far so good.
- Do not expect this page to translate from one language to another (maybe you can do that with Google Translate - kind of .. using cut and paste) . Just know that if you choose a 'voice' other than English all you get is gr-english as read by someone with a very bad foreign accent .. unless that is what you wanted.
- I had to edit the text 'hear' and there so that the result would sound out properly for certain uncommon words .. eg knee-fight vs nephite. Also with scripture references, while (Isaiah 58:13) is read properly as "Isaiah chapter 58 verse 13" less common (LDS) scripture references like (D&C 88: 63) were read as "D and C 88 point 63". Changing it to (D&C 88: 63) ie adding a space improved it slightly to "D and C 88 [slight pause] 63" as a person might read it out loud. Other than that, it did a great job reading back my text (even all the typo's).
Step three: Install and run the audio capture program. I like the free open source Audacity program. On my computer I just find and install it using the Ubuntu Software app or in Terminal I could just type "sudo apt-get install audacity" ..
- with Audacity running, now I need to remap the audio being output from the Chrome ttsreader.com website back into something Audacity can see (or in this case 'hear')
- from Terminal I install and run the pavucontrol helper program. First I use "sudo apt-get install pavucontrol" to install and then just "pavucontrol" to run it.
- Then back in Audacity, I select the new microphone input as "default: Front Mic: 1" (ie the one added by pavucontrol) and now see the sound activity bar moving. Yay.
3Step four: This took a little coordination; but I got the ttsreader page in Chrome ready to play by moving the cursor to the top of the text box and selecting the desired 'voice' and in a second window I got the Audacity program ready to record by opening up a blank workspace, and then I first clicked "record" in Audacity (the red circle icon) and then clicked "play" in ttsreader (the big green right arrow icon). Voila the sweet sound of success!
- the resulting OGG version is about 1/3 the size of the MP3 version (ie 6MB vs 9MB for a 15 minute recording) however, the MP3 format is a bit more supported .. currently.
- This is what I started with: rogerdavies.net/talk-archive/talk-three-sacred-places-homes-chapels-temples-sc-sep-2016
- And this was the result: in a male Mr French version talk-three-sacred-places-sep2016-enm.mp3
- or in a female Siri version talk-three-sacred-places-sep2016-enf.mp3
RSS Feed
