Forum Update - Announcement about May 10th, 2018 update and your password.

Text to Speech

ercoerco Posts: 19,177
Text to Speech chips seem to be dying out. I would think that by now, someone would have a $10 single-chip solution: serial text in, amplified speaker out. Surely the technology exists to do that. But there's no news, no progress. In fact, we're moving backwards; it's hard to find parts anymore. Votrax is dead. Speakjet development boards are backordered most places.

Looks like canned digitized sound is taking over, but I suspect there would be some strong interest in a standalone text-to-speech chip had someone gone down that road.

Anybody know of anything?
"When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

- Pablo Picasso

Comments

  • 20 Comments sorted by Date Added Votes
  • LeonLeon Posts: 7,619
    edited October 2010 Vote Up0Vote Down
    Microchip has free speech synthesis (and recognition) libraries for their dsPIC and PIC32 chips:

    http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=2659

    They would make it quite easy to develop chips that could be used like the dedicated speech chips that used to be available. I've got their software but haven't tried it yet; it compiles OK.
    Leon Heller
    G1HSM
  • electrosyselectrosys Posts: 212
    edited October 2010 Vote Up0Vote Down
    Have you look at: http://www.speechchips.com/
    Actually I have just got my order (very quick shipping)
    I purchased the TTS256 (serial interface) Text to Code IC for SpeakJet.
    There is also a SoundGin Development Board for about the 2/3 of the price!
  • Mike GreenMike Green Posts: 22,744
    edited October 2010 Vote Up0Vote Down
    Chip's Propeller vocal tract simulator works fairly well (here) and PhiPi's phoneme to vocal tract parameter translator (attached) works as a front end. I think someone was working on a text to phoneme translator, but I don't have a reference to that.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 21,966
    edited October 2010 Vote Up0Vote Down
    Mike Green wrote:
    ... and PhiPi's phoneme to vocal tract parameter translator (attached) works as a front end.
    Well, "works" is being rather generous. "Needs work" is more like it. :)

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • ercoerco Posts: 19,177
    edited October 2010 Vote Up0Vote Down
    I figured perhaps the Propeller could save the day, eventually.

    "All in good time, my Pretty!"
    "When you make a thing, a thing that is new, it is so complicated making it that it is bound to be ugly. But those that make it after you, they don’t have to worry about making it. And they can make it pretty, and so everybody can like it when others make it after you."

    - Pablo Picasso
  • RobotWorkshopRobotWorkshop Posts: 2,300
    edited October 2010 Vote Up0Vote Down
    erco wrote: »
    Looks like canned digitized sound is taking over, but I suspect there would be some strong interest in a standalone text-to-speech chip had someone gone down that road.

    Anybody know of anything?

    Hello Erco,

    There is one you may have missed. It is the V8600A module from RC systems:

    http://www.rcsys.com/modules.htm

    They have been around for years and are one of the few survivors from the all companies that have developed/sold speech products. I have a few of their modules and they work great. One nice feature is that they have remained compatible and consistent on the interface throughout the life of that module.

    Robert
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited October 2010 Vote Up0Vote Down
    Phonology standardization is a rather deceptive topic, not at all simple. Traveling as little as 100 miles reveals regional differences in pronunciation. In fact, efforts of trying to standardized a universal English pronunciation pretty much have reached absurd levels of debate. Every English dictionary publisher seems to put forth their own system of pronunciation while the United Nations attempts to enforce the rather exotic Universal Phonetic Alphabet (UPA) to encompass all spoken languages of the world.

    Have you ever heard of "Mid-Atlantic English"? Many people believe it is to become the future of English pronunciation. This is not English pronounced by the the mid-Atlantic states of the USA or any other place on dry land, but is an ideal put forth that is half British and half American as created by the synthesis of rock n' roll between the two countries. And so, rather humorously, they chose to name the dialect for the middle of the Atlantic Ocean.

    So I suppose that any attempt to create a 'speech chip' would best be supported with an IDE that takes samples from the way people really speak and transfers it to text.

    But that is only the first hurdle as spelling is idiosyncratic. So you have to create a vast lexicon that associated each word with the proper phonological elements - regardless of variations in spelling. And there in lies the real dilemma - you have to have a search engine stay ahead of the text it is reading and assemble sentences.

    In some cases, a limited few words can easily be passably spoken. Or idiomatic phrases can be used in a predictable way. But reading whole large texts is not a small microcontroller project.

    In other words, the project requires handling lots and lots of data. As a teacher of English as a foreign language, it is quite embarrassing to present 'standard pronunciation' while often having difficulty understanding Australians and Englishmen.
    Hwang Xian Shen, Puddleby-on-the-Marsh.
    All things considered, I can live and thrive without Microsoft products. LINUX is just fine.
  • kwinnkwinn Posts: 7,873
    edited October 2010 Vote Up0Vote Down
    To come up with a logical and consistent structure for english we should look at the mistakes children make as they are learning to speak. I found that my children and their friends made mistakes in pronunciation, spelling, and grammar that were due to the illogical and exception filled nature of the language. In almost all cases their choices made more sense than what is actually in use.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • HumanoidoHumanoido Posts: 5,770
    edited October 2010 Vote Up0Vote Down
    From time to time the SPO256 speech chip and its companion chip are available. It works on published plans for the BASIC Stamp so you could get it to work with a Propeller chip. I'm using a cache of these until the Propeller chip TTS is perfected.
  • HumanoidoHumanoido Posts: 5,770
    edited October 2010 Vote Up0Vote Down
    Here are the talk demos for the Propeller chip. It seems possible to develop this, using phonemes, into a full TTS.

    http://forums.parallax.com/showthread.php?p=613308
  • LeonLeon Posts: 7,619
    edited October 2010 Vote Up0Vote Down
    I played with those SP0256 chips when they first came out. Another popular speech chip I used around the same time was the Votrax SC=01:

    http://en.wikipedia.org/wiki/Votrax

    It had some advantages over the SP0256, IIRC. I interfaced both of them to my TRS-80.
    Leon Heller
    G1HSM
  • erco wrote: »
    Text to Speech chips seem to be dying out. I would think that by now, someone would have a $10 single-chip solution: serial text in, amplified speaker out. Surely the technology exists to do that. But there's no news, no progress. In fact, we're moving backwards; it's hard to find parts anymore. Votrax is dead. Speakjet development boards are backordered most places.

    Looks like canned digitized sound is taking over, but I suspect there would be some strong interest in a standalone text-to-speech chip had someone gone down that road.

    Anybody know of anything?

    I've been developing a smarter MCU board that is about the same size as an Arduino UNO, but also contains a SpeakJet chip and audio amplifier too ... besides 28 3-pin connectors and other easy to use connections to whatever chip or module you use on the board. "RoboGuts™ S.T.E.A.M. Education Program" http://www.R2Pv1.com/

    RoboGuts™ Intelligent content for 3D printing, a S.T.E.A.M. Education Program http://www.r2pv1.com/

    Experiments to learn how to use various Electronic Components, Structured Computer Programming, Phonemes for Speech &Song in any language, and Art (3D Printing, Finishing, Painting.)

    I have several sound experiments posted and am working on more for the Votrax SC-01 chip and the SPO256-AL2 chip ...

    Right now I'm looking for various schematics for the SPO256-AL2 chip ... I want to find the best working schematic.
  • WhitWhit Posts: 3,929
    edited May 11 Vote Up0Vote Down
    Found this old thread! Fun stuff.

    Have not figured out how to control the volume yet. Here is a little test. This sound in the video is very quiet! But - it works! This uses the code zipped in Mike Greens post above. I used the test code and modified it for the S3.



    Any suggestions Phil?
    Whit+

    "We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
  • WhitWhit Posts: 3,929
    Here is my modified test code. It is called S3_talk_demo_test 1.00
    Whit+

    "We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
  • Whit,

    Try the attached. It's as loud as I can make it without causing overflow in the vocal tract object. Even then, some phrases might cause overflow, which tends to reset the Propeller.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • WhitWhit Posts: 3,929
    Phil,

    I feel like a dunce! All I needed to add was the =9?

    Is that explained anywhere in the comments as regards volume control?

    Works great! Thanks. Thanks. Thanks!
    Whit+

    "We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
  • Whit wrote:
    I feel like a dunce! All I needed to add was the =9?
    Please don't. I also made a change in the talker object. But that's where the danger lies.

    -Phil

    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • WhitWhit Posts: 3,929
    Phil,

    I feel like a dunce! All I needed to add was the =9?

    Is that explained anywhere in the comments as regards volume control?

    Works great! Thanks. Thanks. Thanks!
    Whit+

    "We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
  • Whit wrote:
    Is that explained anywhere in the comments as regards volume control?
    Yes, but it took more than that. I had to modify the talk object to double the volume output. But that's risky, due to the chance of overflow.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • WhitWhit Posts: 3,929
    Thanks so much Phil! I will look at the top object.
    Whit+

    "We keep moving forward, opening new doors, and doing new things, because we're curious and curiosity keeps leading us down new paths." - Walt Disney
Sign In or Register to comment.