diff options
Diffstat (limited to 'client')
| -rw-r--r-- | client/mic.py | 52 | ||||
| -rw-r--r-- | client/stt.py | 23 |
2 files changed, 38 insertions, 37 deletions
diff --git a/client/mic.py b/client/mic.py index ccc322e..2bb8c57 100644 --- a/client/mic.py +++ b/client/mic.py @@ -4,7 +4,8 @@ """ import os -from wave import open as open_audio +import tempfile +import wave import audioop import pyaudio import alteration @@ -81,7 +82,6 @@ class Mic: """ THRESHOLD_MULTIPLIER = 1.8 - AUDIO_FILE = "passive.wav" RATE = 16000 CHUNK = 1024 @@ -155,15 +155,17 @@ class Mic: stream.stop_stream() stream.close() audio.terminate() - write_frames = open_audio(AUDIO_FILE, 'wb') - write_frames.setnchannels(1) - write_frames.setsampwidth(audio.get_sample_size(pyaudio.paInt16)) - write_frames.setframerate(RATE) - write_frames.writeframes(''.join(frames)) - write_frames.close() - - # check if PERSONA was said - transcribed = self.passive_stt_engine.transcribe(AUDIO_FILE, mode=TranscriptionMode.KEYWORD) + + with tempfile.NamedTemporaryFile(mode='w+b') as f: + wav_fp = wave.open(f, 'wb') + wav_fp.setnchannels(1) + wav_fp.setsampwidth(audio.get_sample_size(pyaudio.paInt16)) + wav_fp.setframerate(RATE) + wav_fp.writeframes(''.join(frames)) + wav_fp.close() + f.seek(0) + # check if PERSONA was said + transcribed = self.passive_stt_engine.transcribe(f, mode=TranscriptionMode.KEYWORD) if PERSONA in transcribed: return (THRESHOLD, PERSONA) @@ -175,18 +177,10 @@ class Mic: Records until a second of silence or times out after 12 seconds """ - AUDIO_FILE = "active.wav" RATE = 16000 CHUNK = 1024 LISTEN_TIME = 12 - # user can request pre-recorded sound - if not LISTEN: - if not os.path.exists(AUDIO_FILE): - return None - - return self.active_stt_engine.transcribe(AUDIO_FILE) - # check if no threshold provided if THRESHOLD == None: THRESHOLD = self.fetchThreshold() @@ -226,16 +220,18 @@ class Mic: stream.stop_stream() stream.close() audio.terminate() - write_frames = open_audio(AUDIO_FILE, 'wb') - write_frames.setnchannels(1) - write_frames.setsampwidth(audio.get_sample_size(pyaudio.paInt16)) - write_frames.setframerate(RATE) - write_frames.writeframes(''.join(frames)) - write_frames.close() - - mode = TranscriptionMode.MUSIC if MUSIC else TranscriptionMode.NORMAL - return self.active_stt_engine.transcribe(AUDIO_FILE, mode=mode) + with tempfile.SpooledTemporaryFile(mode='w+b') as f: + wav_fp = wave.open(f, 'wb') + wav_fp.setnchannels(1) + wav_fp.setsampwidth(audio.get_sample_size(pyaudio.paInt16)) + wav_fp.setframerate(RATE) + wav_fp.writeframes(''.join(frames)) + wav_fp.close() + f.seek(0) + mode = TranscriptionMode.MUSIC if MUSIC else TranscriptionMode.NORMAL + transcribed = self.active_stt_engine.transcribe(f, mode=mode) + return transcribed def say(self, phrase, OPTIONS=" -vdefault+m3 -p 40 -s 160 --stdout > say.wav"): # alter phrase before speaking diff --git a/client/stt.py b/client/stt.py index 29fb79c..4a46e71 100644 --- a/client/stt.py +++ b/client/stt.py @@ -33,7 +33,7 @@ class AbstractSTTEngine(object): return True @abstractmethod - def transcribe(self, audio_file_path, mode=TranscriptionMode.NORMAL): + def transcribe(self, fp, mode=TranscriptionMode.NORMAL): pass class PocketSphinxSTT(AbstractSTTEngine): @@ -107,7 +107,7 @@ class PocketSphinxSTT(AbstractSTTEngine): config['dictd_music'] = profile['pocketsphinx']['dictd_music'] return config - def transcribe(self, audio_file_path, mode=TranscriptionMode.NORMAL): + def transcribe(self, fp, mode=TranscriptionMode.NORMAL): """ Performs STT, transcribing an audio file and returning the result. @@ -116,12 +116,17 @@ class PocketSphinxSTT(AbstractSTTEngine): PERSONA_ONLY -- if True, uses the 'Persona' language model and dictionary MUSIC -- if True, uses the 'Music' language model and dictionary """ + decoder = self._decoders[mode] - wavFile = file(audio_file_path, 'rb') - wavFile.seek(44) + fp.seek(44) - decoder = self._decoders[TranscriptionMode.NORMAL] - decoder.decode_raw(wavFile) + # FIXME: Can't use the Decoder.decode_raw() here, because + # pocketsphinx segfaults with tempfile.SpooledTemporaryFile() + data = fp.read() + decoder.start_utt() + decoder.process_raw(data, False, True) + decoder.end_utt() + result = decoder.get_hyp() with open(self._logfiles[mode], 'r+') as f: if mode == TranscriptionMode.KEYWORD: @@ -197,7 +202,7 @@ class GoogleSTT(AbstractSTTEngine): config['api_key'] = profile['keys']['GOOGLE_SPEECH'] return config - def transcribe(self, audio_fp, mode=TranscriptionMode.NORMAL): + def transcribe(self, fp, mode=TranscriptionMode.NORMAL): """ Performs STT via the Google Speech API, transcribing an audio file and returning an English string. @@ -205,11 +210,11 @@ class GoogleSTT(AbstractSTTEngine): Arguments: audio_file_path -- the path to the .wav file to be transcribed """ + url = "https://www.google.com/speech-api/v2/recognize?output=json&client=chromium&key=%s&lang=%s&maxresults=6&pfilter=2" % ( self.api_key, "en-us") - with open(audio_file_path, 'rb') as f: - data = f.read() + data = fp.read() try: headers = {'Content-type': 'audio/l16; rate=%s' % GoogleSTT.RATE} |