17 lines
1.5 KiB
JSON
17 lines
1.5 KiB
JSON
{
|
|
"cid": "",
|
|
"uri": "at://did:plc:vzsvtbtbnwn22xjqhcu3vd6y/ai.syui.log.chat/s55utv52t3rf6",
|
|
"value": {
|
|
"$type": "ai.syui.log.chat",
|
|
"author": "did:plc:vzsvtbtbnwn22xjqhcu3vd6y",
|
|
"content": "あと、可愛い声がなかなか見つからなかった。本当ならgcloudで統一するのが良かったんだけど、なかったから、elevenlabsを使ってる。tokenはかなり高め。でもかなり自然だけどね。\nvmcで口を動かしたり、体を動かしたりする操作は、ぎこちなくて調整が必要。あるいは、パターンを用意しておいて、発話の際はそのどれかを選択するようにするとか。でも、mcpなので結構遅れちゃうんだよね。",
|
|
"createdAt": "2026-01-21T11:34:08.483Z",
|
|
"parent": "at://did:plc:6qyecktefllvenje24fcxnie/ai.syui.log.chat/yznvxcj5bjuhq",
|
|
"root": "at://did:plc:vzsvtbtbnwn22xjqhcu3vd6y/ai.syui.log.chat/vr72pvlhuxnf5",
|
|
"translations": {
|
|
"en": {
|
|
"content": "Also, finding a decent voice synthesiser took longer than expected. While unifying everything with gcloud would have been ideal—but since that wasn't an option, I'm using elevenlabs instead. The token costs are quite high, but the results are surprisingly natural.\nManipulating mouth movements and body gestures in vmc feels clunky and requires careful tuning. Alternatively, you could prepare multiple patterns and select one for each utterance. But even then, the MCP latency makes it somewhat unsatisfactory."
|
|
}
|
|
}
|
|
}
|
|
} |