SpeechFlair
Background
SpeechFlair is a subsidiary of Scicon R&D Inc.
www.sciconrd.com who has been developing speech science technologies for
over 30 years.
Their products are now used in some the most prestigious universities and institutions
worldwide as tools for speech science research.
PCquirerX, MacquirerX, PtchWorks, SynthWorks
are the flagships of speech science tools with numerous publications done in various
disciplines.
Based on 30 years of speech science research and innovation,
SpeechFlair uses some the most sophisticated science technologies to
calculate Pitch and Spectrogram.
How does it work?
Pitch or Intonation
is the vibration of the vocal cords as one speaks, in units of Hz (Hetz or cycles
per seconds). By varying the pitch one can deliver more information and naturalness
to speech.
Many politicians, actors and public speakers hire speech therapists to consult them
using their voice effectively. Now you can do the same at home. Generally, for men
ranges between 70 to 150 Hz and for women 100 to 300 Hz. For men “the deep voice”
which is known to be very effective and attractive, is on the lower range of around
50 to 70 Hz.
James Earl Jones famous voice “I am your father”.
Spectrogram, also
known as “voice print” is the three dimensional plot of frequency, time and
intensity of speech. It delivers virtually all the speech content in one image.
Spectrogram is as unique to individuals as their finger print.
Law enforcement use similar technologies to distinguish and verify individuals by
their voice print signature.
SpeechFlair combines
these two technologies into one simple to use interface for the masses to use without
any prior knowledge of speech science. A model speech is presented for you to listen
to and try to duplicate as close as you can.
Pitch is the easiest one to achieve. Simply vary your pitch pattern closely match
that of he model.
Spectrogram is by far more complicated and difficult. With pitch your voice mimics
the varying intonation pattern and you deliver the same information, but NOT SOUND
like model. If you match the spectrogram, then you also “Sound “like the model.
Have you ever heard some comedians who speak like some other personal and sound
exactly like them? They match both pitch and spectrogram of their models.
Pitch Matching Score uses a proprietary mathematical and statistical modeling formulism to
estimate how close the two tracks are.
It evaluates Pitch only and does not consider any of the formants or other speech variants.
You may actually “hum” the pattern and achieve a good result but it does not reflect in the spectrogram.
It is an estimation and not an exact value.
The result is highly gender dependent and you can maximize the result by varying different parameters.
There are three methods of scoring:
Beginner: approximates the general profile without considering any timing or missing frames.
Intermediate: estimates the profile, considering all the frames without any timing.
Advanced: evaluates the profile, considering all the frames with timing.
How does it benefit me?
Everyone who speaks can benefit.
Autism, children with Autism who lack verbal skills, can greatly benefit
from the visual feedback where they can actually see their responses on screen and
focus on the image as oppose to audio alone.
SpeechFlair provide
a platform where the parents, siblings and friends can be a direct advocate and
share in their teachings and fun with them all at the same time.
Stuttering, sufferers have great difficulty controlling their rate of speech
and starting words.
SpeechFlair technology
provides a simple way where they can follow a timing pattern and speed to control
their speech.
The main process of teaching them to control their speech so far has been with hand
waving and simply telling them to slow down without much success.
SpeechFlair can now
assist them in any environment at any time.
Public speaking is a major fear for many. Most people do not have any confidence
in their ability to guide others with only their voice. A talent can be thought
and learned. SpeechFlair
provides the tools for visual feedback for positive reinforcement.
Senior housing and retirement homes are a great area for this technology
to thrive. As one ages, the speech suffers the most adverse effect. Many institutions
now have on hand speech therapist to help and alleviate this pattern.
SpeechFlair can now be of a tool of choice for round a click assistance
in that endeavor.
Speech pathologist now have the ability to extend their skills way after
their sessions and have their clients practice the lessons one their own, speeding
up the progress. Furthermore, SpeechFlair will provide a chronological history of
the treatment whereby the actual progress/regress can be measured and recalled later.
Pathologists, teachers and others can now publish their models to share with others.
Actors/Actresses now have the ability to practice their lines with a direct
feedback, visualizing the effect of their speech on others.
Singers can now use this technology to practice their voice skills, monitoring
their pitch and its variant (Vibrato) at any time and place.
Politicians routinely hire speech pathologist to assist them in their speech
delivery. Now they can use SpeechFlair as a tool to practice on their own.
School Districts currently use many hours of speech therapist to assist special
needs students for their speech. Now they have an additional tool they can use to
expand their knowledge and expertise way beyond the boundaries of the classroom
and have parent directly get involved with their children advancement.
Learning a new language is a major challenge. Simply repeating the words
on a screen with the current arcane technologies are not satisfactory. Many languages
depend very heavily on intonation pattern and its variation. Changing the intonation
pattern of sentence produces different meaning.
SpeechFlair provides that feedback when one can directly see and correct
the proper pattern.
Who can upload model files?
Anyone!
In fact, we do encourage everyone to participate and upload as many models files in many languages.
The model files have two components, one audio and one labels describing what the audio says.
Use any available audio program, your phone, etc… record a model, save as
“wav” formatted audio file.
If you wish to have the label with the audio , then you would need a special FREE software that generates proper audio and labels.
See Manual
Download and install PCquirerSPF (windows),
or
MacquirerSPF (Mac)
Mac users: You may need to go to System
Preferences->Security & Privacy, and allow third parties to install program in
your computer.
Record your audio or open any wav files, activate the “Label” menu, add labels in exactly the proper location of the audio, save.
The labels can then moved to any other location to get the exact matching.
Upload the model… YOU ARE DONE…
Is it FREE?
SpeechFlair is FREE to use for all for basic testing and usage.
However, due to using large audio files and bandwidth requirements, if you choose to use
SpeechFlair on a regular basis, we are asking a small fee to cover the
cost of the storage and bandwidths.
The audio model publishers can share in revenue depending on the hit count of their models very similar to UTube and similar sites.
How do I sign up?
In one simple step, you will redirected to your PayPal account
where you create a subscription plan.
PayPal controls all the financial transactions.
SpeechFlair does not keep or control any financial
data or records.
To cancel the subscription, simple log into your PayPal account and cancel the subscription.