RE: New speech speed-up library available under GPL
- From: "Sina Bahram" <sbahram@xxxxxxxxx>
- To: <programmingblind@xxxxxxxxxxxxx>
- Date: Wed, 10 Nov 2010 14:37:22 -0500
Exactly. Real numbers of comprehension will be lower, of course, but hopefully
not by a lot if we can keep increasing the fuzzy
numbers.
Take care,
Sina
-----Original Message-----
From: programmingblind-bounce@xxxxxxxxxxxxx
[mailto:programmingblind-bounce@xxxxxxxxxxxxx] On Behalf Of Bill Cox
Sent: Wednesday, November 10, 2010 2:14 PM
To: programmingblind@xxxxxxxxxxxxx
Subject: Re: New speech speed-up library available under GPL
Words per minute are a fuzzy number. I've taken Huckleberry Finn and asked
Open Office to tell me how many words the first chapter
has, and then I play the first chapter through TTS systems. By these numbers,
Sina is listening at around 1,500 words per minute
and claiming to comprehend it. However, we're using the same chapter over and
over, and we're not doing scientific measurements of
comprehension. In truth, we're mostly just fooling around, trying to figure
out how to improve speech at high speed, and we don't
need solid numbers for that.
Bill
On Wed, Nov 10, 2010 at 11:28 AM, qubit <lauraeaves@xxxxxxxxx> wrote:
> lol -- good going Sina --
> I listened to Sina's computer during a skype call last week. Indeed
> it is fast -- but he also turns off all punctuation. I could make out
> a good portion of the words, but it might take me a little practice to
> bump up my listening speed.
> Question, are words-per-minute calculated differently when punctuation
> is on? It does significantly slow down the overall speed.
> --le
>
>
> ----- Original Message -----
> From: "Bill Cox" <waywardgeek@xxxxxxxxx>
> To: <programmingblind@xxxxxxxxxxxxx>
> Sent: Tuesday, November 09, 2010 8:20 PM
> Subject: Re: New speech speed-up library available under GPL
>
>
> Wow, that's great! I've been hacking both espeak and
> speech-dispatcher to try and integrate it, but it's taking me longer
> than I'd thought it would. Did you integrate directly into espeak, or
> some other part of the emacspeak sound stack?
>
> As for NVDA, when Alex or I finish integration with espeak, it will
> hopefully only be a matter of time before it shows up in NVDA.
>
> As for speeding up certain vowels vs certain consonants, sonic doesn't
> do any of that. In fact, one of the reasons I prefer espeak with
> sonic speed up is that espeak does try to play with ratios of silence
> and some sounds as you go faster, and it makes espeak sound worse,
> IMO. If you train to understand a synth at high speed, your ear will
> eventually correct for the problems that we all hear initially in high
> speed speech.
>
> The best example is Eloquence. I have long wondered what magical
> algorithms they used to make it sound so good at high speed. The
> answer is... nothing! When I speed up normal speed Eloquence, it
> sounds almost identical to what Eloquence generates directly at high
> speed. This shows that Eloquence is not fine tuning speech as it
> plays faster... it simply plays faster! Actually, they probably did
> tune it a lot to work well at high speed, but all that tuning applies
> to speech at all rates.
>
> So, if you've ever wondered who the fastest speed listener in the
> world is, I can tell you. It's Sina, at least for now. I know that
> because even Eloquence has a fastest rate - 100%. We've sped up some
> Eloquence samples generated at default speed, and Sina can comprehend
> them well past the 100% speed.
>
> Bill
>
> On Tue, Nov 9, 2010 at 1:37 PM, Alex Midence <alex.midencu@xxxxxxxxx> wrote:
>> He succeeded wonderfully. Even got emacspeak with espeak to work
>> without chopping off the ends of syllables in links and line endings.
>> Also can navigate by char now. In orca and speak-up, espeak now
>> sounds fluid when sped up to about 300 words per minute. As for the
>> clarity of it or the asthetics of the voice, it is still espeak. It
>> doesn't sound as nice as eloquence or dectalk or anything like that
>> but, it's free and its fast and it works just fine. Nice work, Bill.
>>
>> Alex M
>>
>> On 11/9/10, Ken Perry <whistler@xxxxxxxxxxxxx> wrote:
>>> I think Bill is applying this library to the speech dispatcher so
>>> that should take care of making speeding up espeak sound better.
>>>
>>> Ken
>>>
>>> -----Original Message-----
>>> From: programmingblind-bounce@xxxxxxxxxxxxx
>>> [mailto:programmingblind-bounce@xxxxxxxxxxxxx] On Behalf Of Homme,
>>> James
>>> Sent: Tuesday, November 09, 2010 7:01 AM
>>> To: programmingblind@xxxxxxxxxxxxx
>>> Subject: RE: New speech speed-up library available under GPL
>>>
>>> Hi,
>>> It would be wonderful if someone would have a look at Espeak. I
>>> don't know C, so I can't help with that project.
>>>
>>> Jim
>>>
>>> Jim Homme,
>>> Usability Services,
>>> Phone: 412-544-1810. Skype: jim.homme Internal recipients, Read my
>>> accessibility blog. Discuss accessibility here. Accessibility Wiki:
>>> Breaking news and accessibility advice
>>>
>>>
>>> -----Original Message-----
>>> From: programmingblind-bounce@xxxxxxxxxxxxx
>>> [mailto:programmingblind-bounce@xxxxxxxxxxxxx] On Behalf Of Bill Cox
>>> Sent: Monday, November 08, 2010 7:40 PM
>>> To: programmingblind@xxxxxxxxxxxxx
>>> Subject: Re: New speech speed-up library available under GPL
>>>
>>> Hi, Stefik. There aren't any tutorials yet. The project is too new.
>>> I'm currently hacking it into speech-dispatcher to enable Linux
>>> users to enjoy it.
>>>
>>> I do believe it will be useful in a lot of places. Mainly, I was
>>> worried that excellent TTS systems like Eloquence are slowly going
>>> away, leaving people with vision impairments with only slow natural
>>> speech TTS systems. Now I feel that problem is solved, as we can
>>> speed up any natural TTS system and I think achieve very high
>>> listening rates. I've been testing my friends and family (they think
>>> I'm weird, but that's nothing new). My family can listen from 2.5X
>>> to 3X faster than recorded without training, and my old school
>>> friend can listen at 3.5X speed up. I can listen at about 4X, but
>>> I've been training to speed-listen for over a year. All of us can
>>> listen to real voices at faster rates than we can alisten to
>>> Eloquence. This leads me to be hopeful about the future of TTS.
>>>
>>> Bill
>>>> Any tutorials available for how you would use it with various
>>>> existing speech architectures? This sounds like a useful project.
>>>>
>>>> Stefik
>>>>
>>>> On Mon, Nov 8, 2010 at 3:36 PM, Alex Midence
>>>> <alex.midence@xxxxxxxxx>
>>> wrote:
>>>>> On 11/4/10, Bill Cox <waywardgeek@xxxxxxxxx> wrote:
>>>>>> I've release a new very simple library for speeding up or slowing
>>>>>> down speech. It's primary strength is high quality at greater
>>>>>> than 2X speed up. The code can be checked out with:
>>>>>>
>>>>>> $ git clone git://vinux-project.org/sonic
>>>>>>
>>>>>> There are speech samples in the samples directory. The primary
>>>>>> motivation for this library is to enable low speed speech
>>>>>> synthesizers to play at high speed with high quality. My hope is
>>>>>> that many will be as easy to comprehend as Eloquence, giving us a
>>>>>> lot more options in high speed TTS.
>>>>>>
>>>>>> Bill
>>>>>> __________
>>>>>> View the list's information and change your settings at
>>>>>> //www.freelists.org/list/programmingblind
>>>>>>
>>>>>>
>>>>> __________
>>>>> View the list's information and change your settings at
>>>>> //www.freelists.org/list/programmingblind
>>>>>
>>>>>
>>>> __________
>>>> View the list's information and change your settings at
>>>> //www.freelists.org/list/programmingblind
>>>>
>>>>
>>> __________
>>> View the list's information and change your settings at
>>> //www.freelists.org/list/programmingblind
>>>
>>>
>>> This e-mail and any attachments to it are confidential and are
>>> intended solely for use of the individual or entity to whom they are
>>> addressed. If you have received this e-mail in error, please notify
>>> the sender immediately and then delete it. If you are not the
>>> intended recipient, you must not keep, use, disclose, copy or
>>> distribute this e-mail without the author's prior permission. The
>>> views expressed in this e-mail message do not necessarily represent
>>> the views of Highmark Inc., its subsidiaries, or affiliates.
>>> __________
>>> View the list's information and change your settings at
>>> //www.freelists.org/list/programmingblind
>>>
>>> __________
>>> View the list's information and change your settings at
>>> //www.freelists.org/list/programmingblind
>>>
>>>
>> __________
>> View the list's information and change your settings at
>> //www.freelists.org/list/programmingblind
>>
>>
> __________
> View the list's information and change your settings at
> //www.freelists.org/list/programmingblind
>
> __________
> View the list's information and change your settings at
> //www.freelists.org/list/programmingblind
>
>
__________
View the list's information and change your settings at
//www.freelists.org/list/programmingblind
__________
View the list's information and change your settings at
//www.freelists.org/list/programmingblind
Other related posts:
