For his research project WILSON SO applied his draftsmanship to technology, using drawing to program a vocal synthesiser. This is an account of his experience;
Synthesizer V, a voice synthesizer similar to Vocaloid, was developed by Dreamtonics Co., Ltd. I have experimented with this software to see how it works and have tried to cover All the things she said by t.A.T.u.
First, I chose to use the voicebank Eleanor Forte. I tried to put on the notes on piano keyboard interface and found it is quite simple and easy to use. I can draw the note according to the line on different pitches- If I drew on a wrong note, I could just drag the note to the pitch that I wanted. It all started as “la” when the notes were put on the program, by double-clicking the note, I could change the word that she would sing. The green line on the bottom part is a parameter for different features on the vocal. They have different settings such as tension, pitch deviation, loudness, breathiness and more. By drawing the parameter to tune the vocal when I wanted a specific note to be louder, I drew the green line on the parameter higher to achieve the effect.
On the other side, tuning words I wanted her to sing made me realised how difficult it could be when it came to tuning an artificial vocal. An interesting thing I found when doing the first few lines of the song was that the original line of the song is “All the things she said”, but when I input the word “said”, it sounded a bit off with a difficult accent, therefore, I tried to change it into “sad”, which was similar to “said” [Fig. 6]. I found the result much nicer and more coherent to me. After that line, I processed to the line “This is not enough”. As the word “enough” is two syllables, I separate the words into two notes as “en+nough”. Furthermore, I wanted to make the line sounded more coherent and natural, for the “not” before “enough”, I converted it into phonemes /n aa t/ and put the /t/ with “en”ough and it became “/n aa/-ten-nough” instead of “not-en-nough” [Fig. 7]. For the Loudness parameter, the higher the green on the graph, the more appeal the vocal had, the lower it was, and the softer it got. I listened to the original vocal in the track and tried to imitate the vocal. “This” and “is” are strong and push out the air, therefore, the beginning of “this” will be a sudden loud and returning to the normal loudness (Fig. 7) .
The full cover of the song featuring Eleanor Forte can be accessed HERE