Entering words and sentences into the program will result in GUMI reading the input out loud. The site recommends for usage such as: video narration, answering machine pick-up, and audio-book production such as reading aloud. It is not compactiable with Otomachi Una's "Talk" software.
A speech software to allow GUMI to do narration was discussed on Noboru's Twitter account in October 2012, but no details were given at the time. However, there was later a demo on December 24, 2013, and a release date set for January 16, 2014.
It was developed by the company Animo Limited (株式会社アニモ) and the model featured in the program is by a modeler named Sely. Animo Ltd. developed the "FineSpeech Ver 3" for the speech synthesis engine. With this it is possible, in addition to vocal synthesis of basic phrases, to subtly adjust the accent with the mouse and from audio files recorded by microphone, as well as perform very small and microscopic adjustments. The audio file can then be exported after playback.
The software is capable of...
- Text Input and Speech Synthesis: Enter the words or phrases into the text window. It is possible to enter multiple words or phrases. Click the Synthesis button to compile.
- Speech Synthesis Matching: Import an audio file from a computer to match the accent and intonation in Megpoid Talk.
- Adjustment and Accent: For a more natural sound, adjust the intonation finely by changing the speech time. Simply expand or contract the syllable block, and change the average pitch by using the pitch guide lines in the pitch edit window. It is also possible to simply draw a straight line, as well as freehand pitch drawing with the mouse. The edit can be pronounced by switching the display to the character of Megpoid Talk's "GUMI".
- Speed/Volume of Synthesis: This can be done easily by using the utterance speed, speech pitch, intonation, or volume slider.
- Non-Verbal Input: Megpoid Talk comes with 60 different non-verbal files, such as coughs, sneezing, laughs, yawning, and others.
Exporting and effects
VOCALOID3 VSQx Export: Using the Export function, it is possible to export a VSQx file to be used in the VOCALOID3 Engine.
- User Dictionary Registration: Add any word with a viable reading into the User Dictionary. This is useful if you want a specific pronunciation or less common word.
- Mixing and Effect: Megpoid Talk is equipped with a mixer. It Is possible to add effects to the voice using VST Plugin effect. There is also 11 kinds of volume and recording control. Use up to 4 mixing effects simultaneously.
- Recording Effects:
- Dynamics System Modulation System EQ System
- Compressor Chorus 2 Band EQ
- RMS Compressor 6 Band EQ
- DeEsser Filter
- PitchTime System RoomFX System
- PitchShift (Autotune) Delay
- PitchShiftRT (Autotune) Reverb
- Speech Output: Export the voice in WAV or MP3 format. Changes to the sound are directly reflected.
- Text-Based Speech Synthesis: Because of the combination of editing rules and speech synthesis complications, the engine compensates for both disadvantages. This method of bringing real-time speech synthesis from a large database gives the user flexibility and allows them to generate more natural speech.
|Megpoid Talk introduction||YouTube|
|Megpoid Talk - compilation demo (reprint)||YouTube|
|CLIP STUDIO ACTION × Megpoid Talk||YouTube|
Software activation requires internet connection. When using microphone, make sure the device meets specifications.
- Corresponding OS: (64bit, 32bit) SP2 or later (64bit, 32bit), Windows 7 / Vista Windows 8.1 / 8
- CPU: Celeron Dual Core 2.1 GHz or more
- Memory: 1GB or more
- Free hard disk space: 400MB or more
- Display resolution: 1280 × 1024 pixels or higher
- Other: Sound device
- Megpoid Talk actually comes with a DAW built-in to the software, and has 11 different vocal effects, like delay and pitch tuning.
- Users can import their own audio file and Megpoid Talk will attempt to match the accent and intonation
- Users can export as MP3 as well as WAV, which is very convenient
- It has a talking GUMI model, just like Sato Sasara and VOICEROID does.