Mobile menu

Is speech recognition helpful for transcriptions?
Thread poster: Julia Mojik

Julia Mojik  Identity Verified
Germany
Local time: 15:18
Member (2007)
English to German
+ ...
Aug 11, 2008

Hello,

I am a transcriber and I am wondering if any speech recognition program might be helpful in terms of transfering an audio recording in mp3, for instance, directly into Word from the mp3.

That means if I have a German recording that needs to be transcribed into German. Or French into French.

I already read the product description of Voice pro 11...it claims to do everything...still it doesn't say anything about the quality of the direct transfer into Word. It doesn't mention the match rate between recording and Word document.

Has anybody had any experience with speech recognition in terms of transcription work?

-How good is speech recognition and what technical requirements have to be met?
-Is it also useful if their are several persons talking at the same time on the recording? I mean can it distinguish what was said in that case or is it only useful for recordings of single speakers?
-If a distinction could be done with speech recognition which one could you recommend in terms of being user friendly?
-Where can I get further information about speech recognition that explains it sufficiently to someone who hasn't worked with it yet? The product descriptions only seem to be for people who already know what's it about.

Your help will be much appreciated.

Julia


Direct link Reply with quote
 

Peter Linton  Identity Verified
Local time: 14:18
Member (2002)
Swedish to English
+ ...
Like an audio typist Aug 11, 2008

I have done one transcription job using speech recognition in my case, Dragon Preferred) and it worked well. However, I don't think it will do exactly what you want.

The best way to work out what it can do for you is to relate it to audio typing. The only difference is that you are dictating to a computer, not a human typist. Therefore the same limitations apply.

The key to understanding speech recognition software is to realise that it is like an extremely stupid audio typist. It has no understanding of language. It is purely statistical. It converts the sound of your voice into digits, and then compares those digits with its library of digitised speech and tries to find matches. So you need a powerful computer and 2G memory. Then it works like a fast and reasonably intelligent audio typist.

The key to success is the same as with an audio typist -- you need to enunciate clearly, and make corrections.

Would an audio typist understand several people talking at the same time? No. Someone (you) would have to filter out each voice and dictate it separately. In other words, single speakers only.

User-friendly ? Not really, it is complex software, and it takes a little while to get the hang of it.

German and French -- as far as I know, you can have either German or French (and both with English thrown in), but I believe you would need a separate German and French versions. Even then, I think it would only work if you spoke both languages with pretty well no accent.

In short, there is a learning curve. But the rewards are great. I have been using SR for 5 years, and for me it works extremely well -- very high accuracy, much more pleasant than typing, and in fact increases both the quantity and quality of your work. I use it for everything, such as this message.


Direct link Reply with quote
 

Julia Mojik  Identity Verified
Germany
Local time: 15:18
Member (2007)
English to German
+ ...
TOPIC STARTER
SR - like an audio typist Aug 11, 2008

Dear Peter,

thanks a lot for your detailed information and for the background how to understand the idea of it. It helps a lot!

Since I already have to manage the transcription program with hot keys or foot pedal I wonder if it might be reasonable to use an additional (SR) software. That would be like a studio - I guess. Listening to the audio...speaking into the microphone and making correction to the "typist"...

-Do you work with your SR on written texts besides using it for forum messages?
-Can it be combinded with a TM?

Thanks a lot again.


Direct link Reply with quote
 

Peter Linton  Identity Verified
Local time: 14:18
Member (2002)
Swedish to English
+ ...
Possible problem Aug 11, 2008

I am trying to visualise your scenario, in particular whether your transscription program might somehow interfere with SR, particularly in the use of hot keys or hardware. That might present a problem.

Dragon is said to work with certain voice recorders - check their website.

Dragon has one essential hot key -- the + key on the numeric keypad. That is the default hot key for switching the microphone on or off.

However, Dragon does assume that your headphones are plugged into the PC's audio input/output, and it does use the sound card in your PC for digitising your speech. This enables you both to speak into the microphone and to listen to the results in your headphones. But perhaps your transcription program needs the same sockets and soiund card. That suggests that you might have to switch frequently between your transcription program and Dragon, or that you cannot run two applications using the same hardware. That might be a show-stopper. The transcription I did was from an audio cassette, so I cannot answer your question.

As for your other questions, I use SR for practically everything, not just messages. My main use is indeed when translating documents in Word using TMs (in my case Trados). Works fine. However, Dragon cannot dictate into every program. For example, when dictating into Trados TagEditor, Dragon says itis a "non-standard window", but keeps on working. Sometimes you have to dictate into a Dragon tool called DragonPad, then copy and paste it. Sometimes it refuses to dictate at all.


Direct link Reply with quote
 

Julia Mojik  Identity Verified
Germany
Local time: 15:18
Member (2007)
English to German
+ ...
TOPIC STARTER
possible problem Aug 11, 2008

Peter,
thanks a lot again!


Direct link Reply with quote
 
xxx1279
Local time: 09:18
One way to use SR for transcription Aug 12, 2008

Hi Julia,

You should have no problem transcribing digital audio files while using a speech recognition program, although I don't think SR will do as much of the work as you expect (as Peter explained.)

I do a fair amount of Spanish > English transcription work, sometimes using Dragon Naturally Speaking, and sometimes not. I have my pedal set up to work with a transcription program called Express Scribe that you can download for free online, and I use my mouse or oral commands to turn Dragon Naturally Speaking on and off, etc. I was originally hoping that DNS would allow me to transcribe files almost simultaneously; I thought I would be able to listen to the audio file and interpret simultaneously into my microphone and DNS would type my interpretation. This turned out to be (laughably) far from the truth, because DNS can't work nearly that fast, and I don't enunciate that well when speaking quickly. What I do is more like consecutive interpreting: I listen to a phrase, pause the recording using my foot pedal, and then say my translation into the microphone. Then I make any obvious corrections to the typed translation and press the pedal again to listen to the next little portion of audio. I also thoroughly re-read the entire transcription when I finish, because I've found that I cannot catch all of the errors while I'm actively transcribing, and there will always be a few weird mistakes made by the SR program.


Direct link Reply with quote
 

Julia Mojik  Identity Verified
Germany
Local time: 15:18
Member (2007)
English to German
+ ...
TOPIC STARTER
Does it save any time? Aug 12, 2008

Hi Clare,

thanks a lot for your message!

Is it as you do it saving you time or does it end up to the same time volume when you do transcription work as you described in your message?

Thanks a lot in advance.


Direct link Reply with quote
 
xxx1279
Local time: 09:18
Saves time on some jobs, not others Aug 12, 2008

Sometimes the set-up described above does help me transcribe faster, but it seems to depend on the particulars of the audio I'm working with. I'm not even sure what it is about some recordings that increases my efficiency; all I know is that sometimes I use DNS and I feel like I'm getting things done faster, and sometimes I just get annoyed with it and go back to normal typing. (I always use my pedal.) I've noticed too that I dislike dictating in the morning; it seems like excessive effort to enunciate so well first thing!

It sounds like you are already doing the number one thing you can to improve transcription speed, which is using a pedal to start, stop, and rewind the recording. If you are interested in SR and it won't ruin your budget to buy the program, I'd say to go for it. It's not essential for transcriptionists at this point, but you probably will be able to find some way to use the program to save time, whether it is with transcription, translation, or writing emails. I also think SR will continue to improve dramatically in the coming years, so in my opinion, there will come a point where transcriptionists not using SR won't be able to compete on price. Learning to use these programs now may just be a logical investment in staying current in the industry if you devote a fair portion of time to transcription.

You can also use SR to let your wrists rest if you suffer from carpal tunnel syndrome.


Direct link Reply with quote
 

Tania Samsonova  Identity Verified
Canada
Local time: 09:18
Member (2007)
English to Russian
+ ...
A question Feb 7, 2009

Peter Linton wrote:

I have done one transcription job using speech recognition in my case, Dragon Preferred) and it worked well. However, I don't think it will do exactly what you want.



Peter, thank you for your informative answer! I have a (somewhat stupid) question though: to transcript a record, do you feed it to Dragon as an audio file, or just play it on and let Dragon to pick up sound waves?

Thanks

Tania Samsonova


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Is speech recognition helpful for transcriptions?

Advanced search






Across v6.3
Translation Toolkit and Sales Potential under One Roof

Apart from features that enable you to translate more efficiently, the new Across Translator Edition v6.3 comprises your crossMarket membership. The new online network for Across users assists you in exploring new sales potential and generating revenue.

More info »
Protemos translation business management system
Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!

The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.

More info »



All of ProZ.com
  • All of ProZ.com
  • Term search
  • Jobs