I have some audio files from my mobile phone.  They are .amr files.  I don't know any more about them  i.e. sampling rate and that kind of stuff.

So I'm starting right from scratch.

 

Can anyone advise as to what software, preferably free, open source, I could use to transcribe them?

p.s.  I've been trying and downloaded, installed and tried the free version of Braina.

No good. can't get it going.  First it doesn't understand the 'open sound' their instructions tell you to use (and it is not in the command list) and then you find 'open sound properties' and try to use it and the system tells you that you can't use that because you haven't bought the pro version so you must write it out full length.

So you write it out full length and it says it doesnt' understand it.

I doubt it is going to be any good.

 

I have a friend with a smartphone with a free app on it that understands english and chinese - chinese ! - and turns it into text with excellent results and no training.

So this technology is well advanced.  there must be something out there....   ?

 

 

 

 

-------------------------

Change "No" to "Know" w/KnowBrainer 2020
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ex 1

Thanks for the reply.

Can't read it. Text too tiny. That deliberate or an accident?

Seems a little too wordy for what I asked: Can anyone advise as to what software, preferably free, open source, I could use to transcribe them?

Been using voicebase.com. Pretty awful results. Hoping to get in touch with that 'friend' ( casual acquaintance, really) and that app.

We had to use a JPEG to display part of our command but we were concerned that it might be too large for everyone's page. We just doubled the size of our answer which should suffice. All you have to do is Google for the appropriate free utility. It's just that easy.

-------------------------

Change "No" to "Know" w/KnowBrainer 2020
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ex 1

voicebase.com is a services primarily aimed at extracting marketing and customer service keywords and phrases as opposed to producing accurate transcriptions of long continuous speech recordings. For its primary use case it does quite well. For continuous speech recognition, as you said it's awful. If I recall correctly even speaking in my best style barely gave a recognition rate of 85%.

Out of the box with Dragon I get 98 to 99% accuracy depending upon the version. By the way I get 96% accuracy out of Google voice recognition on my Samsung Note 4 speakerphone. When I made recordings on my Samsung Note 4 speakerphone I still get 98% when I transcribe them on my normal Dragon profile. When I sent one of my best test recordings, made with my highest quality mic in a quiet location, for human transcription to an Asian service I only got 94% accuracy transcription from human service whereas I got 99% accuracy transcribing against my DPI 15 profile.

You get what you pay for.

Phil Schaadt

I still want to know the same thing:

Can anyone advise as to what software, preferably free, open source, I could use to transcribe them?

I infer from the replies that I should use Dragon and i have dragon and it's not working too well right now. Might get better after sufficient training. (two or three weeks?)

But I just came from a page where it says the way Dragon handles transcribing audio files is this clever technique of having yourself listen to the file and then at the same time speak it back into 'dragon's' microphone.

got a laugh out of that.

i'll try this cable thing i installed some time back...

if Dragon is not working well it is very likely due to the low quality of your recordings and nothing's going to fix that
what you refer to in your post above is usually called "Echo Dictation" and is really your only choice if your recordings are poor quality.

If you can wait a few months there will be new applications based on the Google Cloud Speech API. But that's not going to be free either.
https://techcrunch.com/2016/03/23/google-opens-access-to-its-speech-recognition-api-going-head-to-head-with-nuance/  

https://cloud.google.com/speech/  

Perhaps there is something out there that I'm unfamiliar with.. However, I closely follow the speech recognition market, beyond Nuance, professionally and if you are looking to transcribe poor quality recordings, of continuous dictation, with recognition rates of 95% or better you are just plain SOL.

Phil Schaadt

the main issue with dragon is getting it to take input from an audio file. i've started a new thread asking about that. as for this question, this thread, i consider it just about closed with an answer that looks something like: nothing open source or free leaving dragon probably the way to go. i said that in my last post. must stop repeating myself. thanks all for your valuable input.

Dragon doesn't transcribe .amr files. Try this to convert to WAV
http://audio.online-convert.com/convert-to-wav  

I transcribe files in Dragon all the time.

Phil Schaadt

Thanks for that. Always good to know of another option. I've bookmarked it. I used VLC for the tests I've done so far. Seems to work fine. I chose mp3. How do you get Dragon to hear your files? You use this virtual cable thing?

My general approach to finding the details of functionality for a complex program is to RTFM.

It's impossible for us to help you because you have not bothered to respond to our requests to detail what version of Dragon you are using and your profile has not been completed.

In whatever version of Dragon that you have, go to the help menu and type "transcribe" it will show you how to do it for your version of Dragon.

Phil Schaadt

Too easy! Wonderful. None of the googling I did or the questions I asked in various places brought me to that. Thanks. Right there under my nose all the time. No need for any 'cables'.

It is not doing too good a job but that's alright. The source is pretty rough. first attempt. cellphone in a car.

Yep, I understand your reference to RTFM and I'm sorry you're so upset that you have to resort to profanities no matter how well disguised.

Having been around a lot I'm very familiar with hundreds of crappy manuals that will put you sleep and drive you mad and waste thousands of hours of your time.
It's a bit stupid not to go directly to the answer if you can.
I come from the days when the computer room (DEC VAX) had two walls covered in shelves carrying manuals.
I come from the days when software typically came with hard cover manuals, great heavy boxes.

And I've moved with everyone else into the days where google is the universal manual and everyone gets their answers at what would have been blinding speed to us back then.

So I'm not rapt in that, Phil, my friend.

And then you say it is impossible to help: and then go on to provide the simple help that was all I ever needed !

So that wasn't right. You could help and you did.

Hence my version of Dragon was not as important as you've maintained.

However, it is version 13.

And you'd like me to fill out my member profile. Well that's very flattering. I'll do what I can.

Thanks for the help.

As previously stated, it is impossible for us to qualify a detailed answer without knowing exactly what you are using. Please avoid generic terms like Dragon 13. This is why we previously requested filling out your user profile. Try using more informative terms terms like Dragon Professional or  Premium 13. From a transcription point of view, these are 2 very different products. Also note that DPI 15 is notably better at transcribing digital recordings because of the Deep Learning speech engine.

-------------------------

Change "No" to "Know" w/KnowBrainer 2020
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ex 1

what's it going to cost me to upgrade to 15? and would it really be worth it? i haven't settled down to give this one a test yet. it might be fine for my purposes. i might well need to make new recordings, get new recording gear.

i mean testing with what i've used here might be pointless. not valid as a test. because the crappy little cellphone and the ambience were too poor.

a better phone - my smartphone maybe - or something dedicated to recording may be what i'll need in the finish.

so it's a bit early for me to decide to start upgrading software.

i'm just a guy who, cruising around, made some aural notes whilst driving - because what else can you do - and is exploring the chance of getting machine transcription of those notes.

it may be a snap. it may be impossible. it's probably something in between. and just where in between it is will probably depend upon the kind of hardware/software i bring to the party.

seems straightforward enough to me.

If you try to use a knuckle busting adjustable wrench in place of proper tools on your car engine, you may only experience limited success. Dragon works the same way.

We can't tell you what the cost will be to upgrade to DPI 15 because of your insistence of not filling out your profile or letting us know what edition of Dragon you are using. We have told you numerous times that generic terms like “Dragon 13” limit our answers. PLEASE be specific. Are you using Dragon 13 Home Edition, Premium, Professional or Legal. For example, the Premium to DPI 15 Upgrade is $200 while the Dragon Professional 13 to DPI 15 Upgrade is $150. The reason this forum includes a member profile section is so that you don't have to keep repeating, depending on your question, your computer specs, sound system, operating system or version/edition of Dragon.

As to the value of upgrading is concerned… you'll find that information all over this forum but if you would like additional details, begin with our KnowBrainer Dragon 14 Review, which compares Ver. 14 to Ver. 13 and then read the KnowBrainer Dragon 15 Review to compare Ver. 15 to Ver. 14. You are currently 2 versions behind the latest technology. Depending on how important it is to transcribe digital recordings, this should be (pardon the pun) a KnowBrainer. Dragon 13 introduces the Deep Learning engine which is an entirely different recipe from all previous Dragon engines which is ideally suited to transcribing proper digital recordings and even introduces the ability to simultaneously transcribe limited interviews (more than 1 speaker), unique (untrained) speakers and even lectures. No previous version of Dragon can remotely handle that kind of transcription. That's why we call DPI 15 the digital recorder version of Dragon.

-------------------------

Change "No" to "Know" w/KnowBrainer 2020
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ex 1

I think it will help more

 

abrogard,

Regardless of what version of Dragon you're using, the Dragon help will provide you with the answers that you want 98% of the time, 100% if you know how to ask.

All you ever have to do is to say "search Dragon help for [word or phrase] . Since the Dragon help, and I believe that it goes all the way back to DNS 13, is dynamic. That is, it's online and constantly being updated. Trying to search Google for issues related to Dragon will most often send you right back here to the KnowBrainer forum. Google will seldom provide you with the answers you're looking for Re: Dragon. Try saying "search Dragon help for transcribe audio".

 

Addendum: BTW, RTFM is a common expression that has been used over and over again in many contexts any is never intended to insult you or anyone else.. Phil, myself, and others provide you with help on this forum free of charge. We are only trying to help, but the more information we have the better we can help you. I to come from the days where printed manuals were prolific and ubiquitous. Nobody ever read them anyway. Things are little different today with the various search engines and so forth, but the best help comes from the Dragon Help.

Chuck

"Man is the animal that intends to shoot himself out into interplanetary space, after having given up on the problem of an efficient way to get himself five miles to work and back each day." - Bill Vaughn

-------------------------

VoiceComputer: the only global speech interface.

The views, thoughts and opinions expressed in this post are my own and do not reflect those of VoiceTeach LLC.

Chuck Runquist
VoiceComputer technical support

Hi all,
To go back to open source alternatives, how about Mozilla's deep speech?

In our opinion, Dragon is still the only serious game in town. As far as open source speech recognition is concerned... You get what you pay for. When it comes to email, Outlook and Dragon pay for themselves but we also rely heavily on email.

-------------------------

Change "No" to "Know" w/KnowBrainer 2020
Trial Downloads
Dragon/Sales@KnowBrainer.com 
(615) 884-4558 ex 1