Subtitle, Transcribe & Voice Recognition
Thread poster: Halastrana

Halastrana
United States
Jan 5, 2016

Greetings Folks!

I am currently on the hunt for a very specific program. I'm not quite sure if it exists all in one at the moment.

This program would be able to perform the following tasks;

1. Voice Recognition Transcription
2. Time-coded Subtitling
3. Foreign Language Voice Recognition


Does anyone know of a program that can provide these three functions? If anyone could help I would be greatfull beyond measure.

Thank you!


 

José Henrique Lamensdorf  Identity Verified
Brazil
Local time: 13:26
English to Portuguese
+ ...
I don't think you'll find one Jan 5, 2016

If any such program existed, you wouldn't see so many people worldwide making a living from translation and subtitling.

YouTube has such a feature, you may see a demo here, and you may couple it with machine translation. The most hilarious such cases I saw were doing it with institutional videos for dermatology clinics.

Let's take one at a time.


1. Voice recognition

Some fellow translators had to use VR software due to repetitive stress injuries or accidents. Most reported that it took them 1-3 months to calibrate the software to their personal enunciation to get 99% accuracy.

Now think of all the accents that exist in any language, and imagine how non-thinking software subroutines would handle them. Think of any joke about misunderstandings between Aussie or Cockney and standard UK English, and you'll get the point.


2. Time-coded subtitling

Good subtitling is all about extracting the gist from the script, and writing it in the most concise way possible.

Can software do it? I doubt it. How would software analyze a line, decide what's most relevant for the plot, and paraphrase that into the most concise phrase?

Though that YouTube contrivance does a good job in syncing its output, it assumes that every speaker pauses at the right places, which is not necessarily true.


3. Foreign language voice recognition

Strictly speaking, every language has countless micro-variants. Unless a skilled actor is trained to accurately mimic someone else's way of speaking, each individual speaks a different micro-variant.

To illustrate, I studied Italian as a foreign language in high school (in Brazil). My teacher was from Florence. Nevertheless my accent in Italian is visibly from Genoa. Later I learned that every Brazilian who studies Italian at school (and not at home, speaking with parents from some other region) will acquire a Genoese accent.

My guess is that Genoa is an ancient port city, and the Portuguese have been great sailors throughout history. Hence they might have had some influence on the local version of Italian.


So, this is it. If speech capture, automatic translation, and automatic subtitling were technically feasible, no matter how much it cost, the major film studios would have already bought it, making thousands of video translators seek some other endeavor.


 

Max Deryagin  Identity Verified
Russian Federation
Local time: 21:26
Member (2013)
English to Russian
- Jan 6, 2016

Halastrana wrote:

Greetings Folks!

I am currently on the hunt for a very specific program. I'm not quite sure if it exists all in one at the moment.

This program would be able to perform the following tasks;

1. Voice Recognition Transcription
2. Time-coded Subtitling
3. Foreign Language Voice Recognition


Does anyone know of a program that can provide these three functions? If anyone could help I would be greatfull beyond measure.

Thank you!


Hi Halastrana,

If by "voice recognition" you mean DNS-like voice recognition (not the automatic kind), then the answer to your question is WinCAPS. It is a professional subtitle preparation tool that supports embedded Dragon NaturallySpeaking. You can purchase it, rent it on a monthly basis or pay-as-you-go (per hour of real work). It even has house styles for homophones.

Hope this helps.

[Edited at 2016-01-06 07:40 GMT]


 

José Henrique Lamensdorf  Identity Verified
Brazil
Local time: 13:26
English to Portuguese
+ ...
How long will we (video translators/subtitlers) conrtinue working as such? Jan 6, 2016

Max Deryagin wrote:

Hi Halastrana,

If by "voice recognition" you mean DNS-like voice recognition (not the automatic kind), then the answer to your question is WinCAPS. It is a professional subtitle preparation tool that supports embedded Dragon NaturallySpeaking. You can purchase it, rent it on a monthly basis or pay-as-you-go (per hour of real work). It even has house styles for homophones.


Max,

If WinCAPS QU4NTUM really does what they say it does, and costs only £1,500 (peanuts for any Hollywood studio), how long shall we continue doing it in the traditional way?

Is now the time for another paradigm shift, like in the late 1980s, when I began using PageMaker and a laser printer, doing DTP while linotype operators were having their last breath?


 

Max Deryagin  Identity Verified
Russian Federation
Local time: 21:26
Member (2013)
English to Russian
- Jan 6, 2016

José Henrique Lamensdorf wrote:

Max Deryagin wrote:

Hi Halastrana,

If by "voice recognition" you mean DNS-like voice recognition (not the automatic kind), then the answer to your question is WinCAPS. It is a professional subtitle preparation tool that supports embedded Dragon NaturallySpeaking. You can purchase it, rent it on a monthly basis or pay-as-you-go (per hour of real work). It even has house styles for homophones.


Max,

If WinCAPS QU4NTUM really does what they say it does, and costs only £1,500 (peanuts for any Hollywood studio), how long shall we continue doing it in the traditional way?

Is now the time for another paradigm shift, like in the late 1980s, when I began using PageMaker and a laser printer, doing DTP while linotype operators were having their last breath?


Hi José,

I am not sure why it should be a paradigm shift all of a sudden — it's been like this for years now. You just input text using your microphone rather than keyboard, and that is all. The latest version of Dragon NaturallySpeaking is fantastic at picking up your voice, unless you have a really bad accent.


 

José Henrique Lamensdorf  Identity Verified
Brazil
Local time: 13:26
English to Portuguese
+ ...
I wonder whether it's workable Jan 6, 2016

Max Deryagin wrote:

I am not sure why it should be a paradigm shift all of a sudden — it's been like this for years now. You just input text using your microphone rather than keyboard, and that is all. The latest version of Dragon NaturallySpeaking is fantastic at picking up your voice, unless you have a really bad accent.


We've all had our experience with spell checkers accepting typos because the "wrong" words also exist.

I recall that the ancient Word 2.0 spell checker would not accept "nuvens" (= "clouds" in PT), and would suggest correcting it to "nu vens" (= "naked you come" in PT). In these present days of cloud computing, that would close the loop to the old assertion that 60% of the Internet content was porn.
icon_eek.gif

Considering this experience, after the voice recognition software has been painstakingly calibrated for ONE speaker's enunciation, EVERYBODY else will have a really bad accent.

Just envision that software working on a dialog between an Aussie and a Texan.

[Edited at 2016-01-06 13:29 GMT]


 

Jing Nie
China
Local time: 00:26
Member (2011)
English to Chinese
+ ...
Try Youtube Jan 7, 2016

You may upload your video clips to Youtube.

https://support.google.com/youtube/answer/2734796?hl=en

Use automatic captioning
YouTube can use speech recognition technology to automatically create captions. Since these are automatically generated, the quality of the captions may vary from video to video. You can always edit or remove the captions.

Automatic captions are available in English, Dutch, French, German, Italian, Japanese, Korean, Portuguese, Russian, and Spanish.


 


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


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

Subtitle, Transcribe & Voice Recognition

Advanced search







SDL Trados Studio 2017 only €435 / $519
Get the cheapest prices for SDL Trados Studio 2017 on ProZ.com

Join this translator’s group buy brought to you by ProZ.com and buy SDL Trados Studio 2017 Freelance for only €435 / $519 / £345 / ¥63000 You will also receive FREE access to Studio 2019 when released.

More info »
Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »



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