def async_dictate(storage_uri, encoding, sample_rate_hertz, language_code):
"""
Transcribe long audio file from Cloud Storage using asynchronous speech
recognition
Args:
storage_uri URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
"""
client = speech_v1.SpeechClient()
# storage_uri = 'gs://cloud-samples-data/speech/brooklyn_bridge.raw'
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
"encoding": encoding,
}
audio = {"uri": storage_uri}
operation = client.long_running_recognize(config, audio)
logging.info("Waiting for dictation results...")
response = operation.result()
text = []
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
text.append(alternative.transcript)
logging.info("Dictation successful.")
return text
def main():
# See http://g.co/cloud/speech/docs/languages
# for a list of supported languages.
# The language code you speak.
language_code = 'th-TH' # a BCP-47 language tag
client = speech.SpeechClient()
config = types.RecognitionConfig(
encoding=enums.RecognitionConfig.AudioEncoding.LINEAR16,
sample_rate_hertz=RATE,
language_code=language_code)
streaming_config = types.StreamingRecognitionConfig(
config=config,
interim_results=True)
# Initial loop value
rounds = 1
while True:
try:
print('streaming loop :' + str(rounds))
with MicrophoneStream(RATE, CHUNK) as stream:
audio_generator = stream.generator()
# Create request data
requests = (types.StreamingRecognizeRequest(audio_content=content) for content in audio_generator)
# POST data to google cloud speech
responses = client.streaming_recognize(streaming_config, requests)
# Now, put the transcription responses to use.
listen_print_loop(responses)
except Exception as err:
print(err)
rounds += 1
def transcribe_file(speech_file):
"""Transcribe the given audio file."""
from google.cloud import speech_v1
from google.cloud.speech_v1 import enums
from google.cloud.speech_v1 import types
import io
client = speech_v1.SpeechClient()
# [START speech_python_migration_sync_request]
# [START speech_python_migration_config]
with io.open(speech_file, 'rb') as audio_file:
content = audio_file.read()
audio = types.RecognitionAudio(content=content)
config = types.RecognitionConfig(
encoding=enums.RecognitionConfig.AudioEncoding.LINEAR16,
sample_rate_hertz=16000,
language_code='ja-JP')
# [END speech_python_migration_config]
# [START speech_python_migration_sync_response]
response = client.recognize(config, audio)
# [END speech_python_migration_sync_request]
# Each result is for a consecutive portion of the audio. Iterate through
# them to get the transcripts for the entire audio file.
for result in response.results:
# The first alternative is the most likely one for this portion.
print(u'Transcript: {}'.format(result.alternatives[0].transcript))
def local_short_recognition(local_file_path):
"""This function transcribes a local short audio file using speech
recognition.
Parameters
----------
local_file_path : Path to local audio file, e.g. /path/audio.wav.
Returns
-------
"""
set_path()
client = speech_v1.SpeechClient()
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
response = client.recognize(
{
"language_code": "pt-BR",
"sample_rate_hertz": 16000,
"encoding":
enums.RecognitionConfig.AudioEncoding.ENCODING_UNSPECIFIED
}, audio)
for result in response.results:
alternative = result.alternatives[0]
print(u"{}".format(alternative.transcript))
def transcribe_audio_to_text(local_file_path: str):
"""
Transcribe a short audio file using Google synchronous speech recognition
"""
client = speech_v1.SpeechClient()
# The language of the supplied audio
language_code = "en-US"
encoding = enums.RecognitionConfig.AudioEncoding.FLAC
config = {
"language_code": language_code,
"encoding": encoding,
}
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
response = client.recognize(config, audio)
for result in response.results:
for alternative in result.alternatives:
#Capture all transcription alternatives that Google created
f = open('flacs-transcribed/' + local_file_path[6:-4] + 'txt', 'a')
f.write("Transcript : " + alternative.transcript + " ")
return
def sample_recognize(language_code, gcs_uri):
"""Transcribe audio file from Google Cloud Storage with word time offsets"""
# [START speech_transcribe_async_word_time_offsets_gcs_core]
client = speech_v1.SpeechClient()
# language_code = 'en-US'
# gcs_uri = 'gs://cloud-samples-data/speech/brooklyn_bridge.raw'
if isinstance(language_code, six.binary_type):
language_code = language_code.decode('utf-8')
if isinstance(gcs_uri, six.binary_type):
gcs_uri = gcs_uri.decode('utf-8')
sample_rate_hertz = 16000
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
enable_word_time_offsets = True
config = {
'sample_rate_hertz': sample_rate_hertz,
'language_code': language_code,
'encoding': encoding,
'enable_word_time_offsets': enable_word_time_offsets
}
audio = {'uri': gcs_uri}
response = client.recognize(config, audio)
for result in response.results:
alternative = result.alternatives[0]
print('Transcript: {}'.format(alternative.transcript))
for word_info in alternative.words:
print('Word: {}'.format(word_info.word))
print('Word start time: {} seconds, {} nanos'.format(
word_info.start_time.seconds, word_info.start_time.nanos))
print('Word end time: {} seconds, {} nanos'.format(
word_info.end_time.seconds, word_info.end_time.nanos))
def long_running_recognize(args):
"""
Transcribe long audio file from Cloud Storage using asynchronous speech
recognition
Args:
storage_uri URI for audio file in GCS, e.g. gs://[BUCKET]/[FILE]
"""
print("Transcribing {} ...".format(args.storage_uri))
client = speech_v1.SpeechClient()
# Encoding of audio data sent.
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"enable_word_time_offsets": True,
"enable_automatic_punctuation": True,
"sample_rate_hertz": args.sample_rate_hertz,
"language_code": args.language_code,
"encoding": encoding,
}
audio = {"uri": args.storage_uri}
operation = client.long_running_recognize(config, audio)
response = operation.result()
subs = []
for result in response.results:
# First alternative is the most probable result
subs = break_sentences(args, subs, result.alternatives[0])
print("Transcribing finished")
return subs
def sample_long_running_recognize(language_code, local_file_path):
"""Transcribe local audio file asynchronously"""
# [START speech_transcribe_async_core]
client = speech_v1.SpeechClient()
# language_code = 'en-US'
# local_file_path = 'Path to local audio file, e.g. /path/audio.wav'
if isinstance(language_code, six.binary_type):
language_code = language_code.decode('utf-8')
if isinstance(local_file_path, six.binary_type):
local_file_path = local_file_path.decode('utf-8')
sample_rate_hertz = 16000
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
'sample_rate_hertz': sample_rate_hertz,
'language_code': language_code,
'encoding': encoding
}
with io.open(local_file_path, 'rb') as f:
content = f.read()
audio = {'content': content}
operation = client.long_running_recognize(config, audio)
print('Waiting for operation to complete...')
response = operation.result()
for result in response.results:
print('Transcript: {}'.format(result.alternatives[0].transcript))
def recognize(local_file_path):
client = speech_v1.SpeechClient()
# The language of the supplied audio
language_code = "ko-KR"
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = 8000
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
audio_channel_count = 1
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
"encoding": encoding,
"audio_channel_count": audio_channel_count,
}
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
operation = client.long_running_recognize(config, audio)
print(u"Waiting for operation to complete...")
response = operation.result()
for result in response.results:
alternatives = result.alternatives
for alternative in alternatives:
# alternative = result.alternatives[0]
print(u"Transcript: {}".format(alternative.transcript))
print(u"Confidence: {}".format(alternative.confidence))
def sample_recognize(storage_uri, model):
"""
Transcribe a short audio file from Cloud Storage using a specified
transcription model
Args:
storage_uri URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
model The transcription model to use, e.g. video, phone_call, default
For a list of available transcription models, see:
https://cloud.google.com/speech-to-text/docs/transcription-model#transcription_models
"""
client = speech_v1.SpeechClient()
# storage_uri = 'gs://cloud-samples-data/speech/hello.wav'
# model = 'phone_call'
# The language of the supplied audio
language_code = "en-US"
config = {"model": model, "language_code": language_code}
audio = {"uri": storage_uri}
response = client.recognize(config, audio)
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
print(u"Transcript: {}".format(alternative.transcript))
def __init__(self, language, supportedLanguages):
# self.language = best_match(language, supportedLanguages)[0]
self.language = "de"
logging.info("created speech input for language: " + self.language)
if (self.language == None or self.language == "und"):
raise ValueError("Language is not supported")
self.client = speech_v1.SpeechClient()
def transcriber(self) -> str:
"""
Transcribe a short audio file using asynchronous speech recognition
Args:
local_file_path Path to local audio file, e.g. /path/audio.wav
"""
client = speech_v1.SpeechClient()
language_code = self.language_code
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = self.sample_rate_herts
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
encoding = enums.RecognitionConfig.AudioEncoding.AMR
config = {
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
"encoding": encoding,
}
with io.open(self.audio_file.file_location.path, "rb") as f:
content = f.read()
audio = {"content": content}
# response = client.recognize(config, audio) this is for the short files
operation = client.long_running_recognize(config, audio)
response = operation.result()
content = [
result.alternatives[0].transcript for result in response.results
]
return "".join(content)
def sample_recognize(storage_uri):
"""
Transcribe short audio file from Cloud Storage using synchronous speech
recognition
Args:
storage_uri URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
"""
client = speech_v1.SpeechClient()
# storage_uri = 'gs://cloud-samples-data/speech/brooklyn_bridge.raw'
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = 16000
# The language of the supplied audio
language_code = "en-US"
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
"encoding": encoding,
}
audio = {"uri": storage_uri}
response = client.recognize(config, audio)
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
print(u"Transcript: {}".format(alternative.transcript))
def speech_to_text(local_file_path):
client = speech_v1.SpeechClient()
language_code = "en-US"
sample_rate_hertz = 48000
config = {
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
# "encoding": encoding,
}
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
f = sf.SoundFile(local_file_path)
aud_len = len(f) / f.samplerate
if aud_len < 60:
response = client.recognize(config, audio)
else:
dest_name = str(uuid.uuid4()) + '.wav'
upload_blob(bucket_name="patched_video_output",
source_file_name=local_file_path,
destination_blob_name=dest_name)
cloud_uri = 'gs://' + 'patched_video_output/' + dest_name
print(cloud_uri)
audio = {"uri": cloud_uri}
operation = client.long_running_recognize(config, audio)
response = operation.result()
transcripted_text = []
for result in response.results:
alternative = result.alternatives[0]
transcripted_text.append(alternative.transcript)
# print(u"Transcript: {}".format(alternative.transcript))
return transcripted_text
def sample_long_running_recognize(storage_uri):
"""
Transcribe long audio file from Cloud Storage using asynchronous speech
recognition
Args:
storage_uri URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
"""
client = speech_v1.SpeechClient()
enable_word_time_offsets = True
# storage_uri = 'gs://cloud-samples-data/speech/brooklyn_bridge.raw'
# Sample rate in Hertz of the audio data sent
# sample_rate_hertz = 16000
sample_rate_hertz = 48000
# The language of the supplied audio
language_code = "en-US"
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
# encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
"enable_word_time_offsets": enable_word_time_offsets
# "encoding": encoding,
}
audio = {"uri": storage_uri}
operation = client.long_running_recognize(config, audio)
print(u"Waiting for operation to complete...")
response = operation.result()
response_dict = {
"transcript": "",
"word_timestamps": defaultdict(list),
"video_url": "https://www.youtube.com/watch?v=PqMGmRhKsnM"
}
print("response", response.results)
for result in response.results:
alternative = result.alternatives[0]
response_dict["transcript"] += alternative.transcript
print(u"Transcript: {}".format(alternative.transcript))
# Print the start and end time of each word
for word in alternative.words:
print(u"Word: {}".format(word.word))
response_dict["word_timestamps"][word.word.lower()].append(
word.start_time.seconds)
print(u"Start time: {} seconds {} nanos".format(
word.start_time.seconds, word.start_time.nanos))
print(u"End time: {} seconds {} nanos".format(
word.end_time.seconds, word.end_time.nanos))
print(json.dumps(response_dict))
def stt(file, channel, hertz, languageCode):
print('exec stt: ', file)
client = speech_v1.SpeechClient()
config = {
"language_code": languageCode,
"sample_rate_hertz": hertz,
"audio_channel_count": channel
}
targetDir, targetFile = os.path.split(file)
gcsURL = f'gs://{bucketName}/{targetFile}'
audio = {"uri": gcsURL}
try:
# アウトプット用ファイル設定
outputFile = os.path.join(targetDir,
'STT_' + targetFile.replace('.wav', '.txt'))
with open(outputFile, mode='w') as f:
# STT実行
operation = client.long_running_recognize(config, audio)
response = operation.result()
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
print(u'Transcript: {}'.format(alternative.transcript))
f.write('{}\n'.format(alternative.transcript))
print('done stt: ', file)
print('#######')
except Exception as e:
print('stt() is ERROR:')
print(e)
print('#######')
def transcribe(local_file_path):
"""
Transcribe a short audio file using synchronous speech recognition
Args:
local_file_path Path to local audio file, e.g. /path/audio.wav
"""
client = speech_v1.SpeechClient()
# local_file_path = 'resources/brooklyn_bridge.raw'
# The language of the supplied audio
language_code = "te-IN"
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = 16000
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
"encoding": encoding,
}
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
response = client.recognize(config, audio)
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
return (u"{}".format(alternative.transcript))
def sample_long_running_recognize(storage_uri):
"""This function transcribes a cloud long audio file using speech
recognition.
Parameters
----------
storage_uri URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
Returns
-------
"""
set_path()
client = speech_v1.SpeechClient()
audio = {"uri": storage_uri}
operation = client.long_running_recognize(
{
"sample_rate_hertz": 16000,
"language_code": "pt-BR",
"encoding":
enums.RecognitionConfig.AudioEncoding.ENCODING_UNSPECIFIED,
}, audio)
for result in operation.result().results:
alternative = result.alternatives[0]
print(u"{}".format(alternative.transcript))
def sample_long_running_recognize(storage_uri):
"""
Transcribe long audio file from Cloud Storage using asynchronous speech
recognition
Args:
storage_uri URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
"""
client = speech_v1.SpeechClient()
# storage_uri = 'gs://cloud-samples-data/speech/brooklyn_bridge.raw'
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = 44100
# The language of the supplied audio
language_code = "ko-KR"
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
# encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
"encoding": 'FLAC',
"audio_channel_count": 2,
"enable_word_time_offsets": True,
}
audio = {"uri": storage_uri}
operation = client.long_running_recognize(config, audio)
print(u"Waiting for operation to complete...")
response = operation.result()
f = open("./data_output.txt", mode='wt')
cf = open("./data_output.csv", 'w')
wr = csv.writer(cf)
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
print(u"Transcript: {}".format(alternative.transcript))
for word in alternative.words:
lst = []
print(u"Word: {}".format(word.word))
f.write(word.word)
f.write(" ")
lst.append(word.word)
print(u"Start time: {} seconds {} nanos".format(
word.start_time.seconds, word.start_time.nanos))
f.write(str(word.start_time.seconds))
lst.append(word.start_time.seconds)
f.write('\n')
print(u"End time: {} seconds {} nanos".format(
word.end_time.seconds, word.end_time.nanos))
wr.writerow(lst)
f.close()
cf.close()
def speech2text(audio):
client = speech.SpeechClient()
try:
response = client.recognize(config=config, audio=audio)
sents = get_sentences(response)
return sents
except:
return 'Not recognized'
def sample_recognize(file, config={}):
if file == None:
logger.error("please input target image file on first argument.")
return None
elif type(file) not in (str, FileStorage):
logger.error(
"please input target filename or FileStorage on first argument.")
return None
if config is None:
logger.error("please set config argument")
return None
elif type(config) is not dict:
logger.error("argument config need to dict type.")
return None
else:
if config.get("language_code", None) is None:
logger.error("please set config.language_code ")
return None
if config.get("sample_rate_hertz", None) is None:
logger.error("please set config.sample_rate_hertz ")
return None
if config.get("max_alternatives", None) is None:
logger.error("please set config.max_alternatives ")
return None
client = speech_v1.SpeechClient()
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
cfg = {
"language_code": config.get("language_code", None),
"sample_rate_hertz": config.get("sample_rate_hertz", None),
"encoding": encoding,
"max_alternatives": config.get("max_alternatives", None),
}
audio = None
if type(file) is str:
with io.open(file, "rb") as f:
content = f.read()
audio = {"content": content}
elif type(file) is FileStorage:
content = file.stream.read()
audio = {"content": content}
file.stream.close()
strs = []
response = client.recognize(cfg, audio)
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
logger.info(u"Transcript: {}".format(alternative.transcript))
strs.append(alternative.transcript)
return strs
def sample_recognize(filename):
"""
Transcribe a short audio file using synchronous speech recognition
Args:
local_file_path Path to local audio file, e.g. /path/audio.wav
"""
if (filename.endswith(".MOV")): #or .avi, .mpeg, whatever.
mp3_filename = filename[:-4] + ".mp3"
flac_filename = filename[:-4] + ".flac"
monoFlac_filename = "mono" + flac_filename
#subprocess.call(['ffmpeg', '-i', filename, mp3_filename])
#subprocess.call(['ffmpeg', '-i', mp3_filename, '-f', 'flac', flac_filename])
#subprocess.call(['ffmpeg', '-i', flac_filename, '-ac', '1', monoFlac_filename])
os.system("ffmpeg -i " + filename + " " + mp3_filename +
" &> /dev/null")
os.system("ffmpeg -i " + mp3_filename + " -f flac " + flac_filename +
" &> /dev/null")
os.system("ffmpeg -i " + flac_filename + " -ac 1 " +
monoFlac_filename + " &> /dev/null")
else:
pass
if (filename.endswith(".flac") and filename[0:4] == "mono"):
metadata = audio_metadata.load(filename)
sample_frequency = metadata['streaminfo']['sample_rate']
else:
sample_frequency = 44100
client = speech_v1.SpeechClient()
# The language of the supplied audio
language_code = "en-US"
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = sample_frequency
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
#encoding = enums.RecognitionConfig.AudioEncoding.MP3
config = {
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
"encoding": "FLAC",
}
with io.open(monoFlac_filename, "rb") as f:
content = f.read()
audio = {"content": content}
response = client.recognize(config, audio)
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
return "{}".format(alternative.transcript)
def sample_recognize(local_file_path, output_dir, language):
"""
Transcribe a short audio file using synchronous speech recognition
Args:
local_file_path Path to local audio file, e.g. /path/audio.wav
"""
filename = local_file_path.split('/')[-1][:-3] + 'txt'
filepath_text = os.path.join(output_dir,filename)
# if continue_skip and os.path.exists(filepath_text):
# return
client = speech_v1.SpeechClient()
# local_file_path = 'resources/brooklyn_bridge.raw'
# The language of the supplied audio
language_code = language
# Sample rate in Hertz of the audio data sent
sample_rate_hertz = 16000
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"language_code": language_code,
"sample_rate_hertz": sample_rate_hertz,
"encoding": encoding,
"audio_channel_count": 1,
"enable_word_time_offsets": True,
"enable_automatic_punctuation":False
}
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
#print("Starting Call")
response = client.recognize(config, audio)
#print("Call ended")
transcript = ""
if response:
if response.results and response.results[0]:
if response.results[0].alternatives and response.results[0].alternatives[0]:
transcript = response.results[0].alternatives[0].transcript
with open(filepath_text, 'w+', encoding='utf8') as file:
file.write(transcript)
def convert_mp3_to_speech():
latest_audio_file = _get_latest_audio_file()
print("latest_audio_file: ", latest_audio_file)
storage_uri = "gs://us-west3-video-enhancer-bb0ff304-bucket/audio_files/" + latest_audio_file
print("storage_uri: ", storage_uri)
client = speech_v1.SpeechClient()
enable_word_time_offsets = True
sample_rate_hertz = 48000
language_code = "en-US"
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
# encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
"enable_word_time_offsets": enable_word_time_offsets
}
audio = {"uri": storage_uri}
operation = client.long_running_recognize(config, audio)
print("Waiting for operation to complete...")
response = operation.result()
response_dict = {
"transcript": "",
"word_timestamps": defaultdict(list),
"video_url": YOUTUBE_URL
}
print("response from SpeechToText", response.results)
verbose = False
for result in response.results:
alternative = result.alternatives[0]
response_dict["transcript"] += alternative.transcript
for word in alternative.words:
response_dict["word_timestamps"][word.word.lower()].append(
word.start_time.seconds)
if verbose:
print(u"Word: {}".format(word.word))
print(u"Start time: {} seconds {} nanos".format(
word.start_time.seconds, word.start_time.nanos))
print(u"End time: {} seconds {} nanos".format(
word.end_time.seconds, word.end_time.nanos))
json_data_for_search_indexing = json.dumps(response_dict)
print("json_data_for_search_indexing: ", json_data_for_search_indexing)
return json_data_for_search_indexing
def sample_long_running_recognize(request=''):
bucket_name = 'visumm-store'
# S2T
client = speech_v1.SpeechClient()
# read input arguments
if request.args and 'input_filename' in request.args:
input_filename = request.args.get('input_filename')
print('got input_filename: ', input_filename)
else:
print('ERROR: no input_filename was provided. existing')
return
# input filepath as GCS uri
storage_uri = 'gs://' + bucket_name + '/' + input_filename
# S2T config
sample_rate_hertz = 44100
language_code = "en-US"
# Encoding of audio data sent. This sample sets this explicitly.
# This field is optional for FLAC and WAV audio formats.
encoding = enums.RecognitionConfig.AudioEncoding.FLAC
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
"audio_channel_count": 2,
"encoding": encoding,
}
audio = {"uri": storage_uri}
operation = client.long_running_recognize(config, audio)
print(u"Waiting for operation to complete...")
response = operation.result()
full_transcript = ''
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
full_transcript = full_transcript + alternative.transcript + '\n'
print('full_transcript: \n', full_transcript)
# write to file
local_fpath = '/tmp/full_transcript.txt'
with open(local_fpath, 'w') as f:
f.write(full_transcript)
# output filename to store
output_filename = input_filename[0:-5] # remove the .flac extension
output_filename = output_filename + '.txt' # add .txt extension
upload_blob(local_fpath, output_filename)
#sample_long_running_recognize()
def __init__(self):
from google.cloud import speech_v1
from google.cloud.speech_v1 import enums
self.client = speech_v1.SpeechClient()
self.config = {
'model': 'video',
'language_code': 'en-US',
'sample_rate_hertz': 16000,
'encoding': enums.RecognitionConfig.AudioEncoding.LINEAR16,
}
def sample_recognize(self, local_file_path, filename, path):
client = speech_v1.SpeechClient()
language_code = "en-US"
enable_word_time_offsets = True
use_enhanced = True
encoding = enums.RecognitionConfig.AudioEncoding.LINEAR16
config = {
"language_code": language_code,
"encoding": encoding,
"enable_word_time_offsets": enable_word_time_offsets,
"use_enhanced": use_enhanced,
}
with io.open(local_file_path, "rb") as f:
content = f.read()
audio = {"content": content}
swears = [
'f**k', 'shit', 'ass', 'bitch', 'w***e', 'dick', 'fuk', "bitches",
"sexual", "anus", "asshole", 'f*****g'
]
response = client.recognize(config, audio)
ts = []
for result in response.results:
alternative = result.alternatives[0]
print(u"Transcript: {}".format(alternative.transcript))
for j in alternative.words:
for key in swears:
if key in j.word:
start = j.start_time.nanos / (10**
9) + j.start_time.seconds
end = j.end_time.nanos / (10**9) + j.end_time.seconds
print(j.word, start, end)
se = [start, end]
ts.extend(se)
print(ts)
bl = ''
for i in range(0, len(ts), 2):
bl += 'between(t\,{0}\,{1})+'.format(ts[i], ts[i + 1])
if (len(bl) != 0):
bl = bl[:-1]
os.system(
'''ffmpeg -i {0} -max_muxing_queue_size 1024 -c:v copy -af volume=0:enable='{1}' {2}'''
.format(path + filename + ".mp4", bl,
path + filename + "filtered" + ".mp4"))
else:
print(path, filename)
os.chdir("/Users/VAISHNAVI/Desktop/mini/uploadedfiles")
command = "copy {0} {1}".format(filename + ".mp4",
filename + "filtered" + ".mp4")
subprocess.call(command, shell=True)
def speechToText(speakerProfileId, storage_uri):
print(speakerProfileId)
os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = "key.json"
client = speech_v1.SpeechClient()
sample_rate_hertz = 16000
language_code = "en-US"
config = {
"sample_rate_hertz": sample_rate_hertz,
"language_code": language_code,
}
with io.open('/tmp/' + storage_uri, "rb") as f:
content = f.read()
audio = {"content": content}
operation = client.long_running_recognize(config, audio)
response = operation.result()
transcript = ""
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
transcript += alternative.transcript
fireStoreClient = firestore.Client()
docReference = fireStoreClient.collection("enrolledUsers").document(
speakerProfileId).get().to_dict()
userName = docReference["name"]
docReferenceScore = fireStoreClient.collection(
'performanceScore').document(str(userName))
importantWords = [
"welcome", "thank you", "sorry", "apologise", "apologize", "good day",
"nice day", "good morning", "good evening", "good noon", "awesome",
"sweet", "hope", "see you", "bye", "hello", "hi", "please", "sure",
"sort", "sorted", "enjoy", "safe"
]
wordsSpoken = dict()
print(transcript)
for word in importantWords:
if word in transcript:
wordsSpoken[word] = transcript.count(word)
print(wordsSpoken)
try:
word_dct = docReferenceScore.get().to_dict()
for word, freq in wordsSpoken.items():
if word in word_dct.keys():
new_freq = word_dct[word]["frequency"] + freq
docReferenceScore.set({word: {
"frequency": new_freq
}},
merge=True)
else:
docReferenceScore.set({word: {"frequency": freq}}, merge=True)
except:
for word, freq in wordsSpoken.items():
docReferenceScore.set({word: {"frequency": freq}}, merge=True)
def sample_long_running_recognize(storage_path, save_path):
"""
Print start and end time of each word spoken in audio file from Cloud Storage
Args:
storage_path can be URI for audio file in Cloud Storage, e.g. gs://[BUCKET]/[FILE]
or local file path
"""
client = speech_v1.SpeechClient()
# storage_uri = 'gs://cloud-samples-data/speech/brooklyn_bridge.flac'
# When enabled, the first result returned by the API will include a list
# of words and the start and end time offsets (timestamps) for those words.
enable_word_time_offsets = True
# The language of the supplied audio
language_code = "en-US"
encoding = enums.RecognitionConfig.AudioEncoding.FLAC
config = {
"enable_word_time_offsets": enable_word_time_offsets,
"language_code": language_code,
"audio_channel_count": 2,
# "sample_rate_hertz": 8000,
"encoding": encoding,
}
if 'gs://' == storage_path[:5]:
audio = {"uri": storage_path}
else:
with io.open(storage_path, "rb") as f:
content = f.read()
audio = {"content": content}
operation = client.long_running_recognize(config, audio)
print(u"Waiting for operation to complete...")
response = operation.result()
name = storage_path.split('/')[-1].split('.')[0]
with open(save_path + 'google_result_' + name + '.txt', 'w') as f:
for result in response.results:
# First alternative is the most probable result
alternative = result.alternatives[0]
f.write(u"Transcript: {}".format(alternative.transcript) + '\n')
# Write the start and end time of each word
for word in alternative.words:
f.write(u"Word: {}".format(word.word) + '\n')
f.write(u"Start time: {} seconds {} nanos".format(
word.start_time.seconds, word.start_time.nanos) + '\n')
f.write(u"End time: {} seconds {} nanos".format(
word.end_time.seconds, word.end_time.nanos) + '\n')
f.close()
print('analysis finished')
def __init__(self):
self.client = speech_v1.SpeechClient()
self.uploader = Uploader()
self.base_config = {
"encoding":
speech_v1.enums.RecognitionConfig.AudioEncoding.LINEAR16,
"enable_word_time_offsets": True,
"profanity_filter": False,
"model": "default",
"enable_automatic_punctuation": True
}