Web   ·   Wiki   ·   Activities   ·   Blog   ·   Lists   ·   Chat   ·   Meeting   ·   Bugs   ·   Git   ·   Translate   ·   Archive   ·   People   ·   Donate
summaryrefslogtreecommitdiffstats
path: root/speech_gst.py
diff options
context:
space:
mode:
authorGonzalo Odiard <godiard@gmail.com>2012-10-24 12:08:35 (GMT)
committer Gonzalo Odiard <godiard@gmail.com>2012-10-24 12:21:20 (GMT)
commitf26a04fc580bcf7dc4118bcbf1358e9f56a27b33 (patch)
tree083900fdb8392dc5cb988b0f2ed4748db861d214 /speech_gst.py
parentb98e1e9589e1fe8378b52372ded165eebdf16c92 (diff)
Finish port of tts functionality to gtk3 and gst 1.0 - SL #4057
This patch solves SL #3705 too (tts button placement in the toolbar) Signed-off-by: Gonzalo Odiard <gonzalo@laptop.org>
Diffstat (limited to 'speech_gst.py')
-rw-r--r--speech_gst.py36
1 files changed, 15 insertions, 21 deletions
diff --git a/speech_gst.py b/speech_gst.py
index 64d4ee3..9b026e4 100644
--- a/speech_gst.py
+++ b/speech_gst.py
@@ -14,7 +14,7 @@
# along with this program; if not, write to the Free Software
# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
-import gst
+from gi.repository import Gst
import logging
import speech
@@ -24,7 +24,7 @@ _logger = logging.getLogger('write-activity')
def get_all_voices():
all_voices = {}
- for voice in gst.element_factory_make('espeak').props.voices:
+ for voice in Gst.ElementFactory.make('espeak', None).props.voices:
name, language, dialect = voice
if dialect != 'none':
all_voices[language + '_' + dialect] = name
@@ -34,29 +34,23 @@ def get_all_voices():
def _message_cb(bus, message, pipe):
- if message.type == gst.MESSAGE_EOS:
- pipe.set_state(gst.STATE_NULL)
+ if message.type == Gst.MessageType.EOS:
+ pipe.set_state(Gst.State.NULL)
if speech.end_text_cb != None:
speech.end_text_cb()
- if message.type == gst.MESSAGE_ERROR:
- pipe.set_state(gst.STATE_NULL)
+ if message.type == Gst.MessageType.ERROR:
+ pipe.set_state(Gst.State.NULL)
if pipe is play_speaker[1]:
speech.reset_cb()
- elif message.type == gst.MESSAGE_ELEMENT and \
+ elif message.type == Gst.MessageType.ELEMENT and \
message.structure.get_name() == 'espeak-mark':
mark = message.structure['mark']
speech.highlight_cb(int(mark))
def _create_pipe():
- pipe = gst.Pipeline('pipeline')
-
- source = gst.element_factory_make('espeak', 'source')
- pipe.add(source)
-
- sink = gst.element_factory_make('autoaudiosink', 'sink')
- pipe.add(sink)
- source.link(sink)
+ pipe = Gst.parse_launch('espeak name=espeak ! autoaudiosink')
+ source = pipe.get_by_name('espeak')
bus = pipe.get_bus()
bus.add_signal_watch()
@@ -70,8 +64,8 @@ def _speech(speaker, words):
speaker[0].props.rate = speech.rate
speaker[0].props.voice = speech.voice[1]
speaker[0].props.text = words
- speaker[1].set_state(gst.STATE_NULL)
- speaker[1].set_state(gst.STATE_PLAYING)
+ speaker[1].set_state(Gst.State.NULL)
+ speaker[1].set_state(Gst.State.PLAYING)
info_speaker = _create_pipe()
@@ -92,19 +86,19 @@ def play(words):
def pause():
- play_speaker[1].set_state(gst.STATE_PAUSED)
+ play_speaker[1].set_state(Gst.State.PAUSED)
def continue_play():
- play_speaker[1].set_state(gst.STATE_PLAYING)
+ play_speaker[1].set_state(Gst.State.PLAYING)
def is_stopped():
for i in play_speaker[1].get_state():
- if isinstance(i, gst.State) and i == gst.STATE_NULL:
+ if isinstance(i, Gst.State) and i == Gst.State.NULL:
return True
return False
def stop():
- play_speaker[1].set_state(gst.STATE_NULL)
+ play_speaker[1].set_state(Gst.State.NULL)