Phrase edit

In A.I.VOICE Editor, you can register phrases that often appear in fixed sentences and other texts in the phrase dictionary. When a registered phrase appears in the input text, the phrase will be read with the registered reading. (For more information on the phrase dictionary, see “Phrase dictionary”)
* This function is available only in the environment where Japanese voices are installed. Also, the registered phrases are only valid when the Japanese voice is used.
Phrase editing is performed on the phrase edit screen in the tuning area at the bottom of the main screen. The phrase editing screen is displayed by clicking the “Edit phrase” in the tuning area.
The phrase editing screen has two main sections, “Accent Edit Mode” and “Voice Effects Edit Mode”. The “Voice Effects Edit Mode” is further divided into four screens corresponding to the four voice effects editable in the A.I.VOICE Editor: volume, speed, pitch, and inflection. Basic phrase editing, such as reading phrases, raising and lowering accents, and inserting pauses, is performed in the accent edit mode (see “Accent editing screen”). In the voice effects edit mode, voice effects can be specified for portions of phrases (see “Voice effect edit mode”). To move to the editing screen, press the button to the left of the phrase editing screen.

Prepare phrases to be edited

To register a new phrase, read the text out loud on the main screen, then display the phrase to be edited on the edit phrase screen.
To edit a phrase that has already been registered, select the phrase to be edited on the phrase list screen and click the “Edit” button to display it on the phrase edit screen
Phrases displayed on the edit phrase screen are valid until you choose to edit another phrase on the phrase list screen or read the text out loud on the main screen.

Accent editing screen

Accent editing is performed in “Accent edit mode”. This mode is initially selected if the “Accent” button on the left side of the phrase editing screen is not selected, press the “Accent” button to move to “Accent edit mode”

The phrase edit screen displays prosody information as shown below

_images/PhraseEdit.png
  • Accent mark
    The round marks connected by lines that appear on the phrase editing screen are called accent marks. A group of accent marks connected by lines is called an accented phrase. Accent marks indicate the accent of the phrase. Accent marks are displayed in units of moras, which are described below.

    PhraseEdit_Accent
  • Mora
    The kana(reading) that appears below the accent mark is called the mora. The mora represents the reading of the phrase.

    PhraseEdit_Mora
  • Accent mark menu
    The pop-up menu that appears when you right-click on an accent mark or on a mora is called the “accent mark menu”. The following operations can be performed on the phrase being edited from the accent mark menu.
    PhraseEdit_Accent_Menu

Menu

Functions

Edit Reading (Y)

Edit readings by accent phrase. (See “Reading edit”)

Join Accent Word (J)

Combine accented clauses. (See “Combining and splitting accented clauses”)

Split Accent Word (S)

Split accented clauses. (See “Combining and splitting accented clauses”)

Devoice (D)

Directs vowel to be silent. (See “Voicing and silencing of vowels”)

Do not devoice (V)

Directs the voicing of vowels. (See “Voicing and silencing of vowels”)

Automate devoiced (A)

If silent is required, it will be automatically silenced. (See “Voicing and silencing of vowels”)

Delete Accent Word (R)

Accent clause will be removed. (See “Delete accent phrase”)

Reading edit

Phrase readings can be edited on an accent phrase basis. To change the reading of an accented phrase, select “Edit Reading (Y)” from the accent marks menu.
_images/PhraseEdit_Yomi1.png
Enter the new reading in full-width katakana and press the “OK” button.
_images/PhraseEdit_Yomi2.png
A new reading is assigned to the accent phrase.
_images/PhraseEdit_Yomi3.png

Accent adjustment (Standard Japanese)

To adjust the accent, drag the accent mark up or down with the mouse.

PhraseEdit_Accent_Edit

Click the “Convert to Standard dialect” button at the bottom right of the phrase editing screen to convert the accent of the current phrase to Standard Japanese specifications.
* The function adjusts the sentence to conform to the rules of standard Japanese with reference to the accent at the time the button is pressed, so the result may not match the result of Japanese analysis of the same sentence.
_images/PhraseEdit_standard_accent.png

Combining and splitting accented clauses

If simply raising or lowering of accents does not result in natural intonation, more complex accent adjustments can be made by combining and splitting accent phrases.
To combine accent phrases, open the accent marker menu for the first mora of the back accent phrase in the two accented phrases to be combined and select “Join Accent Word (J)”.
_images/PhraseEdit_Accent_Join1.png
Two accent phrases are combined to form one accent phrase.
_images/PhraseEdit_Accent_Join2.png
To split an accented phrase, open the Accent Marks menu for the back mora of the two moras that flank the place in the accent phrase where you want to split it, and select “Split Accent Word (S)”.
_images/PhraseEdit_Accent_Separate1.png
One accent phrase is split into two accent phrases.

Delete accent phrase

To delete an accented phrase, select “Delete Accent Word (R)” from the accent marks menu. However, if there is only one accented phrase in a phrase, it cannot be deleted.

Enter pause

Pauses can be inserted before or after an accent phrase. There are three types of pauses available in phrase editing: short pauses, long pauses, and optional long pauses. The short and long pauses on the edit phrase screen correspond to the short and long pauses in the pause settings (see “The length of the pause”)
Pause insertion is done from the pop-up menu (Pause Menu) that appears when you right-click on the highlighted area between accent phrases.
PhraseEdit_Pause_Menu
  • Insert short pause
    Selecting “Insert Short Pause (S)” from the pause menu inserts a short pause between accent phrases, and a symbol appears on the screen to indicate that a short pause has been inserted.
    PhraseEdit_Pause_Short
  • Insert long pause
    Selecting “Insert Long Pause (L)” from the pause menu inserts a long pause between accent phrases, and a symbol appears on the screen to indicate that a long pause has been inserted.
    PhraseEdit_Pause_Long
  • Insert optional long pauses
    An optional long pause is a pause that can be of any length within a predetermined range. Select “Insert arbitary length pause (A)” from the pause menu to open a dialog box where you can enter a pause length in the range of 80 to 30000 milliseconds (30 seconds), then press the “OK” button.
    PhraseEdit_Pause_Manual1
    An optional long pause is inserted between the accent clauses. In the case of an optional long pause, a number representing the pause length can be seen on the screen below the symbol indicating that a pause has been inserted.
    PhraseEdit_Pause_Manual2
  • Delete pause
    To delete an inserted pause, select “Delete Pause (D)” from the pause menu.

Voicing and silencing of vowels

The A.I.VOICE Editor’s text-to-speech engine automatically determines whether vowels should be voiced or not and reads them out loud. However, if you do not get the expected results, you can instruct the text-to-speech engine to voice or silence the vowels.
  • Direct silencing
    Select “Devoice (D)” from the accent marks menu to display a symbol above the mora to indicate silencing.
    PhraseEdit_Unvoicing
  • Direct voicing
    If you select “Do not devoice (V)” from the accent marks menu, a symbol representing voicing will appear above the mora.
    PhraseEdit_Voicing
To cancel the silent or voicing instruction, select “Automate devoiced (A)” from the accent mark menu. The symbol indicating voicing or silencing will disappear from the top of the mora.
Also, by left clicking on the mora switches the state in the following order: “Devoice”, “Do not devoice”, “Automate devoiced” (→ “Devoice”…).

Voice effect edit mode

In the voice effect edit mode, you can specify voice effects on an accent phrase-by-accent phrase basis. The voice effect edit mode is divided into four screens according to the type of voice effect (“volume”, “speed”, “pitch”, and “intonation”), so please press the button corresponding to the voice effect you wish to edit among the buttons on the left side of the phrase edit screen.
In the voice effect edit mode, unlike the accent edit mode, a line representing the voice effect set is displayed. The horizontal axis indicates the position where the voice effect is set, and the vertical axis indicates the size of the voice effect value. In a phrase without voice effect editing, no voice effect is set, so the line representing the default value (1.00) extends directly across the phrase.
_images/PhraseEdit_Mode_Pitch1.png
After editing the voice effect, a line with different heights will be drawn corresponding to the voice effect value you set.
_images/PhraseEdit_Mode_Pitch2.png
The voice effect can be set for each accent phrase. The voice effect value for an accent phrase for which no voice effect is set follows the value of the previous accent phrase. A square point is displayed at the beginning of the accent phrase with the voice effect set, and the set value is displayed above the line.

Set up voice effects

When you move the mouse cursor over an accent phrase for which you wish to set a voice effect, the voice effect value is displayed along with a thin line. This line and value represent the phonetic effect to be set, and the display changes according to the position of the mouse cursor.
_images/PhraseEdit_Mode_Pitch_Edit1.png
Left-clicking while other options are displayed sets the voice effect for the accent phrase.
_images/PhraseEdit_Mode_Pitch_Edit2.png
The table below shows the range of sound effects that can be set.

Types of voice effect

Minimum value

Initial value

Maximum value

Volume

0.00

1.00

2.00

Speed

0.50

1.00

2.00(*)

Pitch

0.50

1.00

2.00

Intonation

0.00

1.00

2.00

* The maximum speaking speed is 4.0 for master control voice effects, but 2.0 is the maximum in the phrase editing voice effects editing mode. For master control voice effects, see “Voice effect”.
In addition, by enabling “extended input range for voice effect values” in the preferences, the input range for each voice effect value will be extended as follows.

Types of voice effect

Minimum value

Initial value

Maximum value

Volume

0.00

1.00

2.00

Speed

0.10

1.00

5.00

Pitch

0.10

1.00

5.00

Intonation

0.00

1.00

5.00

* The value in red is the value to be expanded.

Set initial values

If you want to return to the default value (1.00) after an accent phrase for which you have set a voice effect, and you have difficulty using the mouse to left-click exactly at the 1.00 position, you can also set the default value from the pop-up menu.
  1. Right-click on the accent phrase for which you want to set the initial value, and a pop-up menu will appear. * At this time, any number of candidate values can be displayed.
    PhraseEdit_Mode_Pitch_Edit3
  2. Select “Set default value (D)” in the pop-up menu to set the default value (1.00) when the accent phrase is right-clicked.
    PhraseEdit_Mode_Pitch_Edit4

Deleting voice effects

To remove a voice effect already set, left-click on the square point at the beginning of the accent phrase.
_images/PhraseEdit_Mode_Pitch_Remove1.png
If you have difficulty clicking on the square point, you can remove the voice effect value from the pop-up menu. Right-click with the candidate value displayed for the accent phrase you wish to remove the voice effect from, and a pop-up menu will appear.
_images/PhraseEdit_Mode_Pitch_Remove2.png
Selecting “Delete Setting value (R)” from the pop-up menu will delete the voice effect for that accent phrase.
To delete voice effects for all accent phrases, right-click with the candidate values displayed and select “Remove all accent word settings (A)” from the pop-up menu. You can delete any accent phrase for which a candidate value is displayed.
_images/PhraseEdit_Mode_Pitch_Remove_All.png

The action of voice effects

When voice effects are set in voice effect edit mode, the final synthesized voice will reflect the result of crossed voice effects set in voice effect edit mode by the voice effects of the master control.
For example, if the speed in the master control is specified as 1.2, and the speech speed specified in the phrase is 1.5, the result will be 1.8.
If the result of the crossed value exceeds the minimum or maximum value, it is automatically corrected to the minimum or maximum value of the master control voice effect.
For example, if the master control speed is specified as 4.0, the result will not be 8.0 even if the speed specified in the phrase is 2.0, but will be the maximum value of 4.0.
This works so you can make overall adjustments with the master control voice effects after phrase editing.

Adjustment of word endings

Phrases which headings end in a sentence-final symbol (”。”, “?”, “!”, “。”, “?”, “!”) can specify how to read the end of the phrase.
PhraseTail

Types of endings for sentences

Normal
This is a plainspoken reading, with the voice descending in pitch at the end of the sentence.
Question
The voice rises in pitch at the end of the phrase.
Calling
The reading is when the voice rises slightly in pitch at the end of the sentence.
It is mainly effective when the “joy” style is selected.
Assertion
This is the reading for a strong tone of voice that ends abruptly at the end of a sentence.
Mainly effective when the “anger” style is selected.
When a phrase not registered in the user phrase dictionary is set in the phrase edit field, if the ending is “。”, “。” it will be “Normal” When “?”, “?” it will be “Question” and when “!”, “!” it will be “Assertion” These will be selected automatically when a phrase not registered in the user phrase dictionary.
* The reading at the end of a sentence has different effects depending on the voice and style. When using voices that do not correspond to sentence endings such as “Question” and or “Assertion” even when the sentence ending is changed it will not correspond. Also, the input text may not produce the desired effect.

Listening to phrases

On the phrase editing screen, you can read out the phrase being edited to check whether the editing result is as expected. Pressing the “Play” button on the phrase editing screen will read out the currently displayed phrase with the intonation being edited. Pressing the “Pause” or “Stop” button during reading will pause or stop the reading.
The A.I.VOICE Editor allows you to specify the phrase playback start position in accented phrases. Before pressing the “Play” button, move the bar indicating the playback start position by left clicking.
_images/PhraseEdit_Play_Start.png

Undo edits to phrases

To discard your edits and start over, click the “Undo” button.
_images/PhraseEdit_Revert.png
All edits to the phrase will be undone.

Delete all pauses in a phrase

To delete a pause in a phrase, press the “Remove pauses” button.
All short pauses, long pauses, and optional long pauses are deleted.
_images/PhraseEdit_RemovePause.png

Text analysis

If you want to reacquire Japanese text analysis results when you change the heading or to make adjustments from the beginning, click the “Text analysis” button.
Performs text analysis on the current heading and displays the results in the adjustment column.
_images/PhraseEdit_Analyze.png
* All previous adjustments to the phrase being edited will be discarded.

Register phrases

In order for the results of phrase editing to be reflected in the speech, the edited phrase must be registered in the phrase dictionary. After editing a phrase, click the “Registration” button to register it in the phrase dictionary.
_images/PhraseEdit_Add.png

Delete phrase

Clicking the “Delete” button on the phrase edit screen will delete the phrase being edited from the phrase dictionary.
_images/PhraseEdit_Remove.png
It is also possible to delete phrases in the phrase list screen.

Change phrase headings

The heading of a phrase can be changed by editing the heading field of the phrase edit.
PhraseEdit_EditTitle
Changing the heading does not change the reading of the phrase, accent, etc. Please make adjustments as necessary.
If you change the heading of a registered phrase and register it as a phrase, the original phrase will not be deleted. If the heading is different, it will be registered as a different phrase.

Create a partially different phrase based on a registered phrase (phrase copying).

The heading change function can be used to reduce work when you want to register a large number of phrases in which only is required for some texts, such as a name or a number.
Example: Normally, when registering multiple phrases such as the following, the entire phrase must be adjusted and registered one sentence at a time.
 「本日、東京は快晴です。」
 「本日、東京は曇天です。」
 「本日、東京は雷雨です。」
Using the “Change Heading” function, you can create a new phrase by first adjusting the phrase “本日、東京は快晴です。” and register it, you can change “快晴” to “曇天”, “雷雨” based on that phrase, and create a new phrase by simply adjusting only that part of the phrase.
In the above example, follow the steps below to make the change.
  1. Adjust and register the original phrase.
    PhraseEdit_CopyPhrase1
  2. The registered phrase is being selected for editing.
    PhraseEdit_CopyPhrase2
    * This step is not necessary if you are continuing from the previous section.
  3. Change the heading.
    PhraseEdit_CopyPhrase3
    * Readings, accents, and other adjustments remain unchanged.
  4. Edit of the reading will be made when necessary.
    PhraseEdit_CopyPhrase4
    See also: Reading edit
  5. When phrase adjustment is complete, register the phrase.
    PhraseEdit_CopyPhrase5
  6. Because the heading is changed, it is registered as a separate item from the original phrase.
    PhraseEdit_CopyPhrase6
    * If the edited heading already exists, it will be overwritten.
* Depending on the nature of the change and the intent of the sentence, other accented phrases may need to be adjusted in order to achieve a natural pronunciation of the phrase as a whole. If necessary, please listen to the audio to confirm that the pronunciation is as intended.