Music and Text-to-Speech
It is amazing how sophisticated text-to-speech has become. While still in the uncanny valley between the human voice and synthesized speech, it has become more naturally musical, albeit amusing to hear how robots try to speak English.
Graham the English gentleman is my favorite.
When Graham stumbles upon a word or phrase he doesn't know, he has a tendency to put the emphasis on the second syllable (syl-LA-ble.) For example, his artificial prosody in 'South Korea' puts too much emphasis on the second syllable of 'Korea' making it sound more like the melodic contour of someone exclaiming 'Oh really!':
"Students in the Middle East, Africa and the United States have the greatest faith in their math skills. Students in Japan, South Korea, Hong Kong and Taiwan have much less self-confidence, though they actually do better on the tests." [Clip]
Listen to how Graham attempts to parse Paul Simon's lyric in Kodachrome.
The word Kodachrome becomes garbled as Ko-DACH-rome, with the first syllable either dropped or inaudible. The pronunciation of the word Kodachrome must be on a downbeat to sound like English.
"Kodachrome, you give us those nice bright colors. You give us the greens of summers. Makes you think all the world's a sunny day, oh yeah! I got a Nikon camera. I love to take a photograph. So Mama, don't take my Kodachrome away." [Clip]
Graham begins it on an upbeat with a dotted note value: So Ma-MA don't take my Ko-DACH-rome away
Mono-syllabic words like 'film' that require more lingual manipulations to pronounce and have to be divided, e.g. fil-lum. In fact, when words are sung they are usually mapped over a rhythm in order to make musical sense.
While sung words are usually intelligible, they are not bound to the same rules for speech. The word 'people' has to be pronounced on a downbeat to make sense. When John Lennon used it in 'Imagine' he pronounced it starting on an upbeat: Peo-PLE. This is essentially 'wrong' and the culture has accepted that pronunciation in the context of that song. It could have easily been sung with 'people' on a down beat. But who knows why Lennon chose that way? Perhaps it was too square sounding.
Graham the English gentleman is my favorite.
When Graham stumbles upon a word or phrase he doesn't know, he has a tendency to put the emphasis on the second syllable (syl-LA-ble.) For example, his artificial prosody in 'South Korea' puts too much emphasis on the second syllable of 'Korea' making it sound more like the melodic contour of someone exclaiming 'Oh really!':
"Students in the Middle East, Africa and the United States have the greatest faith in their math skills. Students in Japan, South Korea, Hong Kong and Taiwan have much less self-confidence, though they actually do better on the tests." [Clip]
Listen to how Graham attempts to parse Paul Simon's lyric in Kodachrome.
The word Kodachrome becomes garbled as Ko-DACH-rome, with the first syllable either dropped or inaudible. The pronunciation of the word Kodachrome must be on a downbeat to sound like English.
"Kodachrome, you give us those nice bright colors. You give us the greens of summers. Makes you think all the world's a sunny day, oh yeah! I got a Nikon camera. I love to take a photograph. So Mama, don't take my Kodachrome away." [Clip]
Graham begins it on an upbeat with a dotted note value: So Ma-MA don't take my Ko-DACH-rome away
Mono-syllabic words like 'film' that require more lingual manipulations to pronounce and have to be divided, e.g. fil-lum. In fact, when words are sung they are usually mapped over a rhythm in order to make musical sense.
While sung words are usually intelligible, they are not bound to the same rules for speech. The word 'people' has to be pronounced on a downbeat to make sense. When John Lennon used it in 'Imagine' he pronounced it starting on an upbeat: Peo-PLE. This is essentially 'wrong' and the culture has accepted that pronunciation in the context of that song. It could have easily been sung with 'people' on a down beat. But who knows why Lennon chose that way? Perhaps it was too square sounding.
Comments