iListen Review and Giveaway

I often do interviews via Skype. A Killer feature would be if iListen would records the conversation and create a transcript for me to use, instead of me having to relive the interview a hundred times over just to retype it.

So it would have to be able to recognize two voices talking in dialog.

That’s one future feature I’d love to see, and I would love that license as well!

    +5

Q: What future feature would you like to see in iListen?
A: I hope to have iListen record my lecture and then create a lecture transcript so that i can refer to what my lecturer says by reading a word document. Much easier than searching through a audio file!! =)

License pls…. =P

(subscribed to comments)     +2

Imaging sitting in a meeting where noone have to take notes…because my sweet little macbook does all the work…

    +1

Having IT talk to me. Text-to-speech. So, if I said “Read http://macapper.com/2007/05/31/ilisten-review-and-giveaway/“, it would read all the content at http://macapper.com/2007/05/31/ilisten-review-and-giveaway/.

(subscribed to comments)     +1

I’d like it to get me a beer, then take out my trash and wash the dishes.

But seriously, It’d be nice if I could control all of the functions of my computer without ever touching the keyboard. (app switching, instant messenger, installing programs)

    +1

I think it should take an MP3 file and convert speech from it to text. It would be useful for people who record college lectures on a digital recorder.

    +2

One feature I would like to see is the ability for iListen to filter out background noise. Finding a quiet environment that you can talk clearly in is difficult at university. If I am unable to touch my keyboard and need to do voice commands, chances are I’m making some sort of necessary noise with my hands, like rustling through a book looking for quotations for an essay. Furthermore, if it could filter out background noise, then it could also be effectively used at lecture (there is always somebody talking near you). I assume the best way of doing this would be to train the program to look for one person’s voice pattern (the professor), rather than just scanning the environment’s noise for all words that it can recognize. This might require the program to make a few passes over the audio file, but it would be worth it if the end result was your lecture recorded in both text and audio clearly and without error OR your essay typed without words that iListen thinks the book said when its page was turned. This would allow for some serious multitasking, I think.

    +2

I’d be interested in a Language Pack for iListen. It could be used to learn a new language by checking the user’s pronunciation.

    +4

Oooh… I like Louis’s suggestion.

In addition, I didn’t notice a feature reviewed by you that might be of interest to a few of us actors out there. If iListen could play back to me what I say in my own voice, it would aid in bettering my line delivery rather quickly.

(subscribed to comments)     +2

Like one the comments above, iListen would be great if it can filter out background noise because most of the time I’m in noisy environments, especially my home with all my dogs and kids running around. This would make an excellent tool in capturing the instructor during lecture, freeing yourself from writing and being able to concentrate fully on the material at hand.

    +1

Foreign language support, speech to text for creating transcripts of prior recordings.. It’s such a pain for me to listen to a recorded lecture multiple times just to get the point. I’d rather just read it in a transcript - or even better, I’d like to have the option to listen to it and read it!

    +1

I produce videos for the internet and would love it if iListen could listen to one of my videos and create a transcript from it.

    +1

These are all great ideas, especially the ability to filter out background noise as this would only add versatility to the product, but my vote would be for template support. In the health care industry, the ability to fill out templates with simple voice commands (along with the ability to navigate trees) could save hours of time and lots of money while cutting down on errors.

    +1

Q: What future feature would you like to see in iListen?

I would love to use it for jotting down quick notes on my Mini without looking up from an article.

    +0

I’d like to thank everyone for the wonderful comments here, and would like to respond to them, if I may…

Arjun asked about transcribing interviews. Unfortunately, this is way beyond the capability of today’s computers. Speech recognition (all programs, not just ours) require three things in order to get decent accuracy: an acoustic model of the environment in which it is to be used, adaptation to the way an individual pronounces things, and the user needs to speak his or her punctuation. Someday it may be possible to have software transcribe interviews, but computers will have to be several times more powerful than they are today in order for them to do so. MIT thinks *maybe* 15 to 20 years.

The same is true of lectures, so “mY” and “nan” are also out of luck, as is Jonathan Bloom in regards to transcribing videos. What you can do, however, is to listen to the recording and re-speak it using a profile you have already trained. This not only overcomes the limitations mentioned above, but also results in a much better translation into text, since interviews and lectures tend to use more “conversational” style of speech.

Andrew asked about it talking back to you. Mac OS X already does this. Here’s how. Select this reply in your browser. Now go to the “Safari” menu (or the “Firefox” menu, if you are using Firefox) and select “Services.” From the sub menu, select “Speech,” then select “Start Speaking Text.”

Kyle said he wanted something that controlled his computer. Your wish is granted. iListen already controls virtually any function of your computer, either as it is or with commands that are easy for you to create yourself (we also sell sets of commands for specific applications on the MacSpeech web site).

Brandon said he wanted something that transcribes from the MP3 format. iListen can’t do that because the MP3 format is optimized for music, and isn’t suitable to get the best results for translation to text. But with our optional TranscriptionPak installed, you CAN transcribe from either an AIFF or WAV file.

Tom & Chris A. made comments about filtering out noise. iListen already does this. It comes with a noise canceling microphone, and the microphone set up procedure (which can be performed at any time) further calibrates the microphone to compensate for virtually any amount of noise in the environment. I demonstrated iListen at an Apple Vendor Fair today in Sacramento and it was very noisy in the room. iListen performed flawlessly.

Louis talked about a Language Pack that would allow iListen to learn a new language via pronunciation. While iListen does have the ability to learn new words, learning a language is very complex. There are literally thousands of algorithms that come into play. Unfortunately, computers aren’t yet powerful enough to be able to adapt the language model on the fly to a different language from what the software was programmed to understand.

Seth mentioned playing back in your own voice. There are many utilities, such as Audacity, that already do this. We make software that converts your speech to text. That is a pretty big nut to crack, so it isn’t likely we will be branching out into such a specialized realm any time soon. (It was a very interesting comment, however!)

nan asked about foreign language support. iListen is available in versions for US, UK, and Australian/New Zealand English, as well as German, Italian, and Spanish.

Rob asked about template support. You got it. Load in any template you want and just start talking. Since iListen allows you to talk wherever you would normally type, this is no problem at all. Doctors and lawyers use iListen like this all the time.

Richard said it would be great to use for jotting down quick notes. This is really what iListen is best at already. Simply choose your favorite note-taking program, including Stickies or TextEdit (we like TopXNotes from Tropic Software, btw), and just start talking!

Chuck Rogers, Chief Evangelist
MacSpeech, Inc.

(subscribed to comments)     +2

Chuck,

I am not talking about an attachment that cancels noise. I want the program to do this flawlessly on its own. The main issue I have is when people are speaking clearly nearby. Background noise can be easily filtered, but a coincedental clear voice in the background? I would love to see this program be able to filter that out. If it can do this flawlessly, then I would love to get this free license! But I highly doubt it could perform at a lecture like I would want. The voice of the professor would likely be filtered out as background noise compared to the voices of students talking close to the microphone. No? I hope this helps you understand more of what I was wanting. I guess I would call it Smart Filtering.

I hope this response doesn’t void the one reply policy…

    +1

Tom (and everyone else):

Software cannot do this on its own. It can only get so far. But if you use a MacSpeech-certified noise-canceling microphone, you get exactly what you want.

Yesterday, I demonstrated iListen to Apple employees here in Sacramento. I had another person right next to me doing the exact same thing. The room was so noisy, people had to shout in order for us to hear them. Yet iListen performed flawlessly.

The problem you are going to have is that you can’t use iListen to transcribe your professor’s lecture. Here’s why:

1). Your professor would not have a trained profile in iListen;

2). Your processor would not be speaking his or her punctuation;

3). Your professor would not be using a noise-canceling microphone.

But if YOU were using iListen with a noise-canceling microphone with a profile YOU have trained and YOU were speaking your punctuation, you would get excellent results with iListen no matter how loud the professor or students were speaking around you.

The primary issue is that absolutely NO software can give you adequate accuracy if the person speaking does not have a trained profile and is not speaking his or her punctuation, regardless of whether they are using a noise-canceling microphone or not. We are literally 15-20 years away from that type of technology being available to us on a personal computer.

Chuck Rogers, Chief Evangelist
MacSpeech, Inc.

(subscribed to comments)     +0

Chuck,

I am not trying to argue about this. The contest was a feature that you would like to see in the future. I would love to see something that could do what my contest entry was (filter out everything but the voice you wanted). The technology might not be here, but it will be in the future. I’m sorry if I offended you in any way, but I am simply competing to win a license for software that I wish to have but cannot afford.

Tom

    +1

Tom:

Absolutely no offense here. I just wanted to make it clear that the technology already does that within the guidelines presented, and what the limitations were that would make technology infeasible in the near future.

It was my understanding the spirit of the question would be for features we could actually introduce. Maybe I got that wrong.

Chuck Rogers, Chief Evangelist
MacSpeech, Inc.

(subscribed to comments)     +0

Chuck,

In that case. It seems like you have all of your bases covered. You have a pretty comprehensive piece of software. As you explained above to everyone, either the feature already exists in the software or OSX or is not technically feasible.

Tom

    +0

I’d love for iListen to provide voice authentication.

That is, I’d love it if I could use iListen to “encrypt” files, manage “revolving” passwords for websites, or lock my computer, and make it only accesible by speech recognition. There’d have be fail-safe (reboot the machine in safe mode or super user mode, and enter cryptic command/code and password).

As an offshot, this may well save me from having to remember a gazillion different passwords — yet how hard is it for a thief to emulate my voice speaking a specific sequence of sounds?

(subscribed to comments)     +0

It might be a bit like ErichD’s idea, but still……

I’d like to be able to have a list of various different clippings of text, and then, when a word is said, the words be typed into a box. This would be very helpful for passwords, but also for saving your hands typing words that you type every day.

Despite that, if this piece of software works better than the other software I have used offering the same function, you guys have done a very, very, very good job!

    +0

I’d like it to do something like TextExpander or work with it so it could insert snippets by voice easily.

    +0

EricD, Alex, and Ben Druker (and everyone else):

iListen already has a feature called “Text Macros” that allows you to do exactly that. With TextMacros you can have iListen type passwords, snippets of text, or form letters - anything from one character to 32,000 characters (that’s about 8 pages of text!) can be typed with one spoken phrase. We even have an option to paste the text in using the clipboard!

We do not have a means of encrypting the passwords, but they are stored within your iListen profile, which would make them very difficult to find unless someone knew where to look.

Chuck Rogers, Chief Evangelist
MacSpeech, Inc.

(subscribed to comments)     +1

Chuck,

I’ll try to re-state what I’d like to see (yea, I know it already has macros…)

LOG-IN requiring VOICE authentication. As a means of security, iListen could hook into other security software out there that would send relvant data (IP address, anyhting else) as soonas possible int he event that a “Hack” occured - indicating that eihter the machine is stolen or currently being broken into.

i.e. something in addition to the encrtyption that already exists ont he Mac. Hell I’d even love it if my keychain, instead of first asking for atpyped password, asked for voice authentication. Then allowed me to type the password.

There’s always these solutions, which seem like a lot of trouble for what should be “Mac”-simple:

http://www.tmcnet.com/biomag/features/roundup030102.htm

Reading a little into that, I could also suggest that the software could be used to “authenticate” users of software or documents (determined by the user) to verify authorship/ownership. For example, if I’m working on a book or say, investigative news piece, there’s a stage when I’d like this document to stay only accesible to me - and not just anyone who could type my password. Not just so that I could prevent links, but also to prevent incrimination.

Hopefully, this clears up the general idea fo what I’m looking for.

On a completely different note (pun intended), there’s an ap form the MDA thing that some fremworks from iListen might be lent to (or maybe not….):

http://www.stratfordisland.com/whistler/

(subscribed to comments)     +1

Please forgive me for the many, many typos above….

(subscribed to comments)     +1

I would like tos ee iListen implement into its program a visualization program that, at the same time as diverting sounds to text, puts that same sound into view, as if through an iTunes sort of visualizer, with chaning modes and symbols and backgrounds…perhaps the words would pop up at the same time. This would be a wonderful way for artists to express themselves.

(subscribed to comments)     +1

I would like iListen to recognize medical terminology.

    -1

Blue tooth enabled! Would be great to use with Jawbone that state-of-the art noise cancelling cool headphone/mic.

(subscribed to comments)     +0