[jaunty karmic] espeak missing words and phonemes from sentences
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
espeak (Ubuntu) |
Fix Released
|
Medium
|
Ubuntu Audio Team | ||
Bug Description
Hello,
A tiny test of Espeak speech synhesizer:
If I run the following command several times, espeak sometime looses the beginning of statement and starts to speak from "help me to find...". I simply repeat the command rather quickly several times.
echo "I beg your pardon, could you help me to find this address." | espeak
Another example
echo "All your base are belong to us. Do you agree?" | espeak -v en-us -p 10 -s 140
The active sound card is:
lspci | grep -i audio
04:02.0 Multimedia audio controller: Creative Labs SB0400 Audigy2 Value
==================
Possible fix/patch:
I got the sources, and noticed in the Makefile:
AUDIO = portaudio
#AUDIO = pulseaudio
#AUDIO = sada
But guess what? ubuntu now uses pulse... I simply changed it to...
#AUDIO = portaudio
AUDIO = pulseaudio
#AUDIO = sada
.. Recompiled.. stumbled (needed to install libpulse-dev)... and Yipee!!!
description: | updated |
Changed in espeak (Ubuntu): | |
importance: | Low → Medium |
tags: | added: patch-needswork |
I can confirm espeak is broken. I've tried on at least 3 computers with various hardware configurations - laptops, desktops, fast, slow, all 32-bit Jaunty.
Mostly, it plays super-duper fast (like, an entire sentence in a fraction of a second).
On two computers, I received the following messages until I uninstalled bluez-alsa. service_ open: connect() failed: Connection refused (111) service_ open: connect() failed: Connection refused (111)
bt_audio_
bt_audio_
And now, it plays a sentence, but it comes out so fast it's like just a tick.
As posted elsewhere on the web, the wav generation appears to work, eg:
#!/bin/sh
espeak -w ~/.espk.wav "$1"
aplay ~/.espk.wav
rm ~/.espk.wav
does work.