Audiobook Plug-in

To download a copy follow the instructions on the Download page.

Several publicly available text-to-speech cloud services are available for use, they typically offer a try-before-you-buy option and generally offer sample access to the service for without cost. Upgrading to a paid version will be necessary when transforming larger documents.

IBM Cloud Services

The IBM Text to Speech service processes text and natural language to generate synthesized audio output complete with appropriate cadence and intonation. It is available in several voices:

Introduction: Getting Started.

Create an instance of the service:

1. Go to the Text to Speech External link icon page in the IBM Cloud Catalog.
2. Sign up for a free IBM Cloud account or log in.
3. Click Create.

Copy the credentials to authenticate to your service instance:

1. From the IBM Cloud dashboard External link icon, click on your Text to Speech service instance to go to the Text to Speech service dashboard page.
2. On the Manage page, click Show to view your credentials.
3. Copy the API Key and URL values.
4. Within the plug-in alter the file cfg/configuration.properties to hold your API Key and URL.

Microsoft Azure

The Speech Services allow you to convert text into synthesized speech and get a list of supported voices for a region using a set of REST APIs. Each available endpoint is associated with a region. A subscription key for the endpoint/region you plan to use is required.

Introduction: Getting Started.

Create an instance of the service:

1. Go to Try Cognitive Services
2. Select the Speech APIs tab.
3. Under Speech Services, select the Get API Key button.
4. Agree to the terms and select your locale from the drop-down menu.
5. Sign in by using your Microsoft, Facebook, LinkedIn, or GitHub account.

You can sign up for a free Microsoft account at the Microsoft account portal. To get started, click Sign in with Microsoft and then, when asked to sign in, click Create one. Follow the steps to create and verify your new Microsoft account.

After you sign in to Try Cognitive Services, your free trial begins. The displayed webpage lists all the Azure Cognitive Services services for which you currently have trial subscriptions. Two subscription keys are listed beside Speech Services. You can use either key in your applications.

Copy the credentials to authenticate to your service instance:

1. Copy either of the API Key and Endpoint values.
2. Within the plug-in alter the file cfg/configuration.properties to hold your API Key and URL.

When running the mp3 or audiobook transforms, the male voice corresponding to the @xml:lang attribute of the root topic will be chosen to render the speech. Use the --ssml.gender=female parameter to switch to a female voice. If no voice of the preferred gender can be found, the default will be used.

A list of available voices can be found within following files:

• cfg/attrs/bing.voice-attr.xsl
• cfg/attrs/watson.voice-attr.xsl

Each listing shows the default male and female voices for a language, plus any regional variants which are available:

<!-- Voices speaking in English -->
<xsl:attribute-set name="__voice__en__male">
    <xsl:attribute name="voice">en-US_MichaelVoice</xsl:attribute>
</xsl:attribute-set>
<xsl:attribute-set name="__voice__en__female">
    <xsl:attribute name="voice">en-US_AllisonVoice</xsl:attribute>
</xsl:attribute-set>
<!-- Voices speaking in Regional English -->
<xsl:attribute-set name="__voice__en-us__female">
    <xsl:attribute name="voice">en-US_AllisonVoice</xsl:attribute>
</xsl:attribute-set>
<!--xsl:attribute-set name="__voice__en-us__female">
    <xsl:attribute name="voice">en-US_LisaVoice</xsl:attribute>
</xsl:attribute-set-->
<xsl:attribute-set name="__voice__en-gb__female">
    <xsl:attribute name="voice">en-GB_KateVoice</xsl:attribute>
</xsl:attribute-set>

As you can see the en-US_AllisonVoice is currently the preferred female voice for all documents marked up as @xml:lang="en" and @xml:lang="en-US".

• to alter the en preferences, replace the text within the <xsl:attribute name="voice"> element with the preferred voice.
• to alter the en-us preferences, comment out the existing selection and uncomment the new preferred voice.

Some DITA tags such as <p> and <b> translate directly to SSML, however there is rich vocabulary of audio effects which are missing from the vanilla DITA specification. These can be accommodated using the @props attribute added to <ph> tag. Examples are given below. The listing is mainly based on the IBM Text to Speech Programming Guide, however the DITA plug-in is not service specific so some additional tags can be used. Obviously common substitutions should be replaced with <keyword> elements for consistency of reuse.

The say-as tag allows the author to indicate information on the type of text contained within the tag and to help specify the level of detail for rendering the text. The required attribute for this tag is interpret-as . There are two optional attributes, format and detail, which are only used with particular values within the interpret-as attribute. These optional attributes are illustrated within the entries for their associated values.

• letters: This value spells out the characters in a given word within the enclosed tag.

<ph props="say-as interpret-as(letters)">Hello</ph>

• digits: This value spells out the digits in a given number within the enclosed tag.

<ph props="say-as interpret-as(digits)">123456</ph>

• vxml:digits: This value performs the same function as the digits value.

<ph props="say-as interpret-as(vxml:digits)">123456</ph>

• date This value will speak the date within the enclosed tag, using the format given in the associated format attribute. The format attribute is required for use with the date value of interpret-as, but if format is not present, the engine will still attempt to pronounce the date.

<ph props="say-as interpret-as(date) format(mdy)">12/17/2005</ph>
<ph props="say-as interpret-as(date) format(ymd)">2005/12/17</ph>
<ph props="say-as interpret-as(date) format(dmy)">17/12/2005</ph>
<ph props="say-as interpret-as(date) format(ydm)">2005/17/12</ph>
<ph props="say-as interpret-as(date) format(my)">12/2005</ph>
<ph props="say-as interpret-as(date) format(md)">12/17</ph>
<ph props="say-as interpret-as(date) format(ym)">2005/12</ph>

• ordinal - This value will speak the ordinal value for the given digit within the enclosed tag.

<ph props="say-as interpret-as(ordinal)">2</ph>
<ph props="say-as interpret-as(ordinal)">1</ph>

• cardinal - This value will speak the cardinal number corresponding to the Roman numeral within the enclosed tag.

Super Bowl <ph props="say-as interpret-as(cardinal)">XXXIX</ph>

• number - This value is an alternative to using the values given above. Using the format attribute to determine how the number is to be interpreted, you can enter one series of number and have it pronounced several different ways, as in the example. The example also includes two different ways of pronouncing a series of numbers as a telephone number. To have the series pronounced with the punctuation included, you must add the detail attribute.

<ph props="say-as interpret-as(number)">123456</ph>
<ph props="say-as interpret-as(number) format(ordinal)">123456</ph>
<ph props="say-as interpret-as(number) format(cardinal)">123456</ph>
<ph props="say-as interpret-as(number) format(telephone)">555-555-5555</ph>
<ph props="say-as interpret-as(number) format(telephone) detail(punctuation)">555-555-5555</ph>

• vxml:boolean - This value will speak yes or no depending on the value given within the enclosed tag.

<ph props="say-as interpret-as(vxml:boolean)">true</ph>
<ph props="say-as interpret-as(vxml:boolean)">false</ph>

• vxml:date - This value works like the date value, except that the format is predefined as YYYYMMDD. When a value is not known, or you do not wish it to be displayed, a question mark is used to replace that value, as shown in the example.

<ph props="say-as interpret-as(vxml:date)">20050720</ph>
<ph props="say-as interpret-as(vxml:date)">????0720</ph>
<ph props="say-as interpret-as(vxml:date)">200507??</ph>

• vxml:currency - This value is used to control the synthesis of monetary quantities. The string must be written in the UUUmm.nn format, where UUU is the three character currency indicator specified by ISO standard 4217, and mm.nn is the amount.

<ph props="say-as interpret-as(vxml:currency)">USD45.30</ph>

If there are more than two decimal places in the number within the enclosed tag, the amount will be synthesized as a decimal number followed by the currency indicator. If the three character currency indicator is not present, the number will be synthesized as a decimal only, with no pronunciation of currency type.

<ph props="say-as interpret-as(vxml:currency)">USD45.329</ph>

• vxml:phone - This value will speak a phone number with both digits and punctuation, similar to the number value used with format(telephone).

<ph props="say-as interpret-as(vxml:phone)">555-555-5555</ph>

The SSML phoneme tag enables users to provide a phonetic pronunciation for the enclosed text. This tag has two attributes:

• alphabet - This attribute specifies the phonology used. The supported alphabets to designate are ipa for the International Phonetic Alphabet, and ibm for the SPR representation.

• ph - This attribute specifies the pronunciation. It is a required attribute. This example shows how a pronunciation for "tomato" is specified using the IPA phonology, where the symbols are given using Unicode:

<ph props="phoneme alphabet(ipa) ph(t&#x259;mei&#x27E;ou&#x325;)">tomato</ph>

This example shows how a pronunciation for "tomato" is specified using the SPR phonology:

<ph props="phoneme alphabet(ibm) ph(.0tx.1me.0fo)">tomato</ph>

This tag is used to indicate that the text included in the alias attribute is to replace the text enclosed within the tag when speech is synthesized. The only attribute for this tag is the alias attribute, and it is required.

<ph props="sub alias(International Business Machines)">IBM</ph>

This tag is used when a change in voice is required. Although all attributes listed are optional, without any attributes defined an error will result. The optional attributes are:

• age Accepted values are positive integers between the ages of 14 and 60 for both male and female.
• gender Accepted values are male and female.
• name Accepted values are the installed voices’ names.
• variant Accepted values are positive integers.

Examples:

<ph props="voice age(60)">Sixty year-old's voice.</ph>
<ph props="voice gender(female)">This is a female voice.</ph>
<ph props="voice name(Allison)">Use the IBM TTS voice named Allison.</ph>
<ph props="voice name(Allison, Andrew, Tyler)">Use the first available IBM TTS voice named in the given list.</ph>

The <emphasis> element requests that the contained text be spoken with emphasis (also referred to as prominence or stress).

• level: the optional level attribute indicates the strength of emphasis to be applied. Defined values are strong, moderate, none and reduced. The default level is moderate. The meaning of strong and moderate emphasis is interpreted according to the language being spoken (languages indicate emphasis using a possible combination of pitch change, timing changes, loudness and other acoustic differences). The reduced level is effectively the opposite of emphasizing a word. For example, when the phrase "going to" is reduced it may be spoken as "gonna". The none level is used to prevent the synthesis processor from emphasizing words that it might typically emphasize.

That is a <ph props="emphasis"> big </ph> car!
That is a <ph props="emphasis level(strong)"> huge </ph>bank account!

Emphasis can also be achieved using the <b> tag.

That is a <b> big </b> car!
That is a <b props="level(strong)"> huge </b>bank account!

This tag inserts pauses into the spoken text. It has the following optional attributes:

• strength - This attribute specifies the length of a pause in terms of varying strength values: none,x-weak,weak,medium,strong, or x-strong.
• time - This attribute specifies the length of the pause in terms of seconds or milliseconds. The values formats are NNNs for seconds or NNNms for milliseconds.

Different sized <ph props="break strength(none)"/>pauses.
Different sized <ph props="break strength(x-weak)"/>pauses.
Different sized <ph props="break strength(weak)"/>pauses.
Different sized <ph props="break strength(medium)"/>pauses.
Different sized <ph props="break strength(strong)"/>pauses.
Different sized <ph props="break strength(x-strong)"/>pauses.
Different sized <ph props="break time(1s)"/>pauses.
Different sized <ph props="break time(1000ms)"/>pauses.

This tag controls the pitch, range, speaking rate, and volume of the text. all attributes are optional, but if no attribute is given an error results.

Here is a description of the optional attributes:

• pitch - This attribute modifies the baseline pitch for the text enclosed within the tag. Accepted values are either:, a number followed by the Hz designation, a relative change, x-low, low, medium, high, x-high, default

• range This attribute modifies the pitch range for the text enclosed within the tag. Accepted values for this attribute are the same as the accepted values for pitch.

• rate - This attribute indicates a change in the speaking rate for contained text. Accepted values are: - a relative change - a positive number, x-slow, slow, medium, fast, x-fast, default

The rate is specified in terms of words-per-minute. If the speaking rate is 50 words per minute, then rate=50. If the setting is rate=+10, the speaking rate will be 10 words per minute faster than your current rate setting.

• volume - This attribute modifies the volume for the contained text. The range for values is 0.0 to 100.0 or the relative values of : silent, x-soft, soft, medium, loud, x-loud, default

Examples:

<ph props="prosody pitch(150Hz)"> Modified pitch </ph>
<ph props="prosody pitch(-20Hz)"> Modified pitch </ph>
<ph props="prosody pitch(+20Hz)"> Modified pitch </ph>
<ph props="prosody pitch(-12st)"> Modified pitch </ph>
<ph props="prosody pitch(+12st)"> Modified pitch </ph>
<ph props="prosody pitch(x-low)"> Modified pitch </ph>
<ph props="prosody range(150Hz)"> Modified pitch range</ph>
<ph props="prosody range(-20Hz)"> Modified pitch range</ph>
<ph props="prosody range(+20Hz)"> Modified pitch range</ph>
<ph props="prosody range(-12st)"> Modified pitch range</ph>
<ph props="prosody range(+12st)"> Modified pitch range</ph>
<ph props="prosody range(x-high)"> Modified pitch range</ph>
<ph props="prosody rate(slow)"> Modified speaking rate</ph>
<ph props="prosody rate(+25)"> Modified speaking rate</ph>
<ph props="prosody rate(-25)"> Modified speaking rate</ph>
<ph props="prosody volume(88.9)">Modified volume</ph>
<ph props="prosody volume(loud)">Modified volume</ph>

This tag inserts recorded elements into the generated audio. The only attribute is src and is required. This attribute specifies the location of the file to be inserted.

<ph props="audio src(http://www.myfiles.com/files/beep.wav)"/>