Text-to-Speech (TTS) Synthesis

Try out some online text-to-speech synthesizers.  Think about what types of text words and phrases the synthesizer is likely to produce incorrectly and WHY.  In class we discussed seven different types of potential errors; your notes from this lecture would be a good place to start.  Remember that the more fancy or high-end the synthesizer, the less likely it will probably be to make mistakes.  So try some humdrum synthesizers as well.   Turn in the following  filled out.  

You will be trying three different text strings.  Each should test for a different type of error.  Try each of the text strings on three different synthesizers and report whether they produced the string accurately or not.  If not, describe precisely the nature of the error.  Comment also on the naturalness of the synthesis.

Here are some sites to get started, but you needn’t limit yourself to these if you find others (that are free and do not require downloading an app).  These will generate an audio file that you will need to play and listen to on your computer.   (Note that while there are many downloadable tools one could install to do TTS, for this class homework we can only accept TTS systems that can be used online [i.e. without any downloading or installation requirements]).

Copy and paste the template below into a file and add your responses.  Make sure that the links are working (clickable) hyperlinks.  (Help in keyboarding IPA).  Please make sure the filename starts with your last name, e.g. SmithSally_TTSHW.pdf and that your name is on the actual document as well.   Turn in your work as a file to the class assignment dropbox (Core 103) or via email if you are an FSEM student. 

Text String 1

Exact Text String Typed:

State why you chose this string to test—what type of error might be expected and/or why do you think a synthesis failure or inaccuracy might occur.

Synthesizer 1

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices): 

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressiv?

Synthesizer 2

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice  specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Synthesizer 3

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Text String 2

Exact Text String Typed:

State why you chose this string to test—what type of error might be expected and/or why do you think a synthesis failure or inaccuracy might occur.

Synthesizer 1

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Synthesizer 2

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Synthesizer 3

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Text String 3

Exact Text String Typed:

State why you chose this string to test—what type of error might be expected and/or why do you think a synthesis failure or inaccuracy might occur.

Synthesizer 1

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Synthesizer 2

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?

Synthesizer 3

Exact Website address:

Name of synthesizer (if more than one at this website):

Parameter settings and/or voice specified for synthesizer (if there are any choices):

Result for intelligibility & accuracy--If deficient, describe precisely why the production was difficult to understand, inaccurate or inappropriate.

Result for naturalness--Did the speech sound natural/human/expressive?