一開始是用espeak,很容易使用,但是聽起來太像機器人了。
voice在/usr/share/espeak-data下
espeak -f File 就可以讀出File的內容。速度,音量也都很好設定,man espeak就知道了。
現在正在試festival,發音比espeak好多了,但是男聲有點混濁。
festival –tts File 可以讀出File的內容。
voice在/usr/share/festival/voice中
命令 festival> (voice.list) ;列出voices
設定檔.festivalrc
設定速度,在.festivalrc中加入
(Parameter.set ‘Audio_Command “aplay -q -c 1 -t raw -f s16 -r $(($SR*140/100)) $FILE”)
The 140/100 means 140% of original speed which seems about right to me for most texts.
設定聲音,在.festivalrc中加入
(set! voice_default ‘voice_us1_mbrola)
測試,以festival播今天的日期和時間
echo `date ‘+%A %B %d %H %M %S’`|festival –tts
轉到wave檔
text2wave -o output.wav text.to.speak.txt -eval “(voice_us1_mbrola)”
###待測試之其他語音軟體
Simon Flexible speech recognition software
Julius Two-pass large vocabulary continuous speech recognition engine
Jovie Text-to-speech system for KDE
Orca Screen Reader Scriptable screen reader
Praat Speech Analyser Sofware for speech analysis and synthesis