msb-public
/
PaddleSpeech
mirror of https://github.com/PaddlePaddle/PaddleSpeech
1
0
Code Issues Projects Releases Wiki Activity

modify, test=doc

Browse Source
pull/1554/head
lym0302 4 years ago
parent 87ec33a647
commit 8ef92a9495
7 changed files with 79 additions and 96 deletions
Show Stats Download Patch File Download Diff File

4
demos/speech_server/cls_client.sh
Unescape View File

@ -0,0 +1,4 @@
#!/bin/bash
wget -c https://paddlespeech.bj.bcebos.com/PaddleAudio/zh.wav https://paddlespeech.bj.bcebos.com/PaddleAudio/en.wav
paddlespeech_client cls --server_ip 127.0.0.1 --port 8090 --input ./zh.wav --topk 1

2
docs/source/released_model.md
Unescape View File

@ -69,7 +69,7 @@ GE2E + FastSpeech2 | AISHELL-3 |[ge2e-fastspeech2-aishell3](https://github.com/
Model Type | Dataset| Example Link | Pretrained Models | Static Models Model Type | Dataset| Example Link | Pretrained Models | Static Models
:-------------:| :------------:| :-----: | :-----: | :-----: :-------------:| :------------:| :-----: | :-----: | :-----:
PANN | Audioset| [audioset_tagging_cnn](https://github.com/qiuqiangkong/audioset_tagging_cnn) | [panns_cnn6.pdparams](https://bj.bcebos.com/paddleaudio/models/panns_cnn6.pdparams), [panns_cnn10.pdparams](https://bj.bcebos.com/paddleaudio/models/panns_cnn10.pdparams), [panns_cnn14.pdparams](https://bj.bcebos.com/paddleaudio/models/panns_cnn14.pdparams) | [panns_cnn6_static.tar.gz](https://paddlespeech.bj.bcebos.com/cls/inference_model/panns_cnn6_static.tar.gz), [panns_cnn10_static.tar.gz](https://paddlespeech.bj.bcebos.com/cls/inference_model/panns_cnn10_static.tar.gz), [panns_cnn14_static.tar.gz](https://paddlespeech.bj.bcebos.com/cls/inference_model/panns_cnn14_static.tar.gz) PANN | Audioset| [audioset_tagging_cnn](https://github.com/qiuqiangkong/audioset_tagging_cnn) | [panns_cnn6.pdparams](https://bj.bcebos.com/paddleaudio/models/panns_cnn6.pdparams), [panns_cnn10.pdparams](https://bj.bcebos.com/paddleaudio/models/panns_cnn10.pdparams), [panns_cnn14.pdparams](https://bj.bcebos.com/paddleaudio/models/panns_cnn14.pdparams) | [panns_cnn6_static.tar.gz](https://paddlespeech.bj.bcebos.com/cls/inference_model/panns_cnn6_static.tar.gz)(18M), [panns_cnn10_static.tar.gz](https://paddlespeech.bj.bcebos.com/cls/inference_model/panns_cnn10_static.tar.gz)(19M), [panns_cnn14_static.tar.gz](https://paddlespeech.bj.bcebos.com/cls/inference_model/panns_cnn14_static.tar.gz)(289M)
PANN | ESC-50 |[pann-esc50](../../examples/esc50/cls0)|[esc50_cnn6.tar.gz](https://paddlespeech.bj.bcebos.com/cls/esc50/esc50_cnn6.tar.gz), [esc50_cnn10.tar.gz](https://paddlespeech.bj.bcebos.com/cls/esc50/esc50_cnn10.tar.gz), [esc50_cnn14.tar.gz](https://paddlespeech.bj.bcebos.com/cls/esc50/esc50_cnn14.tar.gz) PANN | ESC-50 |[pann-esc50](../../examples/esc50/cls0)|[esc50_cnn6.tar.gz](https://paddlespeech.bj.bcebos.com/cls/esc50/esc50_cnn6.tar.gz), [esc50_cnn10.tar.gz](https://paddlespeech.bj.bcebos.com/cls/esc50/esc50_cnn10.tar.gz), [esc50_cnn14.tar.gz](https://paddlespeech.bj.bcebos.com/cls/esc50/esc50_cnn14.tar.gz)
## Punctuation Restoration Models ## Punctuation Restoration Models

159
paddlespeech/server/bin/paddlespeech_client.py
Unescape View File

@ -70,13 +70,10 @@ class TTSClientExecutor(BaseExecutor):
choices=[0, 8000, 16000], choices=[0, 8000, 16000],
help='Sampling rate, the default is the same as the model') help='Sampling rate, the default is the same as the model')
self.parser.add_argument( self.parser.add_argument(
'--output', '--output', type=str, default=None, help='Synthesized audio file')
type=str,
default="./output.wav",
help='Synthesized audio file')
def postprocess(self, response_dict: dict, outfile: str) -> float: def postprocess(self, wav_base64: str, outfile: str) -> float:
wav_base64 = response_dict["result"]["audio"] #wav_base64 = response_dict["result"]["audio"]
audio_data_byte = base64.b64decode(wav_base64) audio_data_byte = base64.b64decode(wav_base64)
# from byte # from byte
samples, sample_rate = soundfile.read( samples, sample_rate = soundfile.read(
@ -93,37 +90,38 @@ class TTSClientExecutor(BaseExecutor):
else: else:
logger.error("The format for saving audio only supports wav or pcm") logger.error("The format for saving audio only supports wav or pcm")
duration = len(samples) / sample_rate
return duration
def execute(self, argv: List[str]) -> bool: def execute(self, argv: List[str]) -> bool:
args = self.parser.parse_args(argv) args = self.parser.parse_args(argv)
try: input_ = args.input
url = 'http://' + args.server_ip + ":" + str( server_ip = args.server_ip
args.port) + '/paddlespeech/tts' port = args.port
request = { spk_id = args.spk_id
"text": args.input, speed = args.speed
"spk_id": args.spk_id, volume = args.volume
"speed": args.speed, sample_rate = args.sample_rate
"volume": args.volume, output = args.output
"sample_rate": args.sample_rate,
"save_path": args.output
}
st = time.time()
response = requests.post(url, json.dumps(request))
time_consume = time.time() - st
response_dict = response.json()
duration = self.postprocess(response_dict, args.output)
try:
time_start = time.time()
res = self(
input=input_,
server_ip=server_ip,
port=port,
spk_id=spk_id,
speed=speed,
volume=volume,
sample_rate=sample_rate,
output=output)
time_end = time.time()
time_consume = time_end - time_start
response_dict = res.json()
logger.info(response_dict["message"]) logger.info(response_dict["message"])
logger.info("Save synthesized audio successfully on %s." % logger.info("Save synthesized audio successfully on %s." % (output))
(args.output)) logger.info("Audio duration: %f s." %
logger.info("Audio duration: %f s." % (duration)) (response_dict['result']['duration']))
logger.info("Response time: %f s." % (time_consume)) logger.info("Response time: %f s." % (time_consume))
return True return True
except BaseException: except Exception as e:
logger.error("Failed to synthesized audio.") logger.error("Failed to synthesized audio.")
return False return False
@ -136,7 +134,7 @@ class TTSClientExecutor(BaseExecutor):
speed: float=1.0, speed: float=1.0,
volume: float=1.0, volume: float=1.0,
sample_rate: int=0, sample_rate: int=0,
output: str="./output.wav"): output: str=None):
""" """
Python API to call an executor. Python API to call an executor.
""" """
@ -151,20 +149,11 @@ class TTSClientExecutor(BaseExecutor):
"save_path": output "save_path": output
} }
try: res = requests.post(url, json.dumps(request))
st = time.time() response_dict = res.json()
response = requests.post(url, json.dumps(request)) if not output:
time_consume = time.time() - st self.postprocess(response_dict["result"]["audio"], output)
response_dict = response.json() return res
duration = self.postprocess(response_dict, output)
print(response_dict["message"])
print("Save synthesized audio successfully on %s." % (output))
print("Audio duration: %f s." % (duration))
print("Response time: %f s." % (time_consume))
print("RTF: %f " % (time_consume / duration))
except BaseException:
print("Failed to synthesized audio.")
@cli_client_register( @cli_client_register(
@ -193,24 +182,27 @@ class ASRClientExecutor(BaseExecutor):
def execute(self, argv: List[str]) -> bool: def execute(self, argv: List[str]) -> bool:
args = self.parser.parse_args(argv) args = self.parser.parse_args(argv)
url = 'http://' + args.server_ip + ":" + str( input_ = args.input
args.port) + '/paddlespeech/asr' server_ip = args.server_ip
audio = wav2base64(args.input) port = args.port
data = { sample_rate = args.sample_rate
"audio": audio, lang = args.lang
"audio_format": args.audio_format, audio_format = args.audio_format
"sample_rate": args.sample_rate,
"lang": args.lang,
}
time_start = time.time()
try: try:
r = requests.post(url=url, data=json.dumps(data)) time_start = time.time()
# ending Timestamp res = self(
input=input_,
server_ip=server_ip,
port=port,
sample_rate=sample_rate,
lang=lang,
audio_format=audio_format)
time_end = time.time() time_end = time.time()
logger.info(r.json()) logger.info(res.json())
logger.info("time cost %f s." % (time_end - time_start)) logger.info("Response time %f s." % (time_end - time_start))
return True return True
except BaseException: except Exception as e:
logger.error("Failed to speech recognition.") logger.error("Failed to speech recognition.")
return False return False
@ -234,15 +226,9 @@ class ASRClientExecutor(BaseExecutor):
"sample_rate": sample_rate, "sample_rate": sample_rate,
"lang": lang, "lang": lang,
} }
time_start = time.time()
try: res = requests.post(url=url, data=json.dumps(data))
r = requests.post(url=url, data=json.dumps(data)) return res
# ending Timestamp
time_end = time.time()
print(r.json())
print("time cost %f s." % (time_end - time_start))
except BaseException:
print("Failed to speech recognition.")
@cli_client_register( @cli_client_register(
@ -270,22 +256,19 @@ class CLSClientExecutor(BaseExecutor):
def execute(self, argv: List[str]) -> bool: def execute(self, argv: List[str]) -> bool:
args = self.parser.parse_args(argv) args = self.parser.parse_args(argv)
url = 'http://' + args.server_ip + ":" + str( input_ = args.input
args.port) + '/paddlespeech/cls' server_ip = args.server_ip
audio = wav2base64(args.input) port = args.port
data = { topk = args.topk
"audio": audio,
"topk": args.topk,
}
time_start = time.time()
try: try:
r = requests.post(url=url, data=json.dumps(data)) time_start = time.time()
# ending Timestamp res = self(input=input_, server_ip=server_ip, port=port, topk=topk)
time_end = time.time() time_end = time.time()
logger.info(r.json()) logger.info(res.json())
logger.info("Response time %f s." % (time_end - time_start)) logger.info("Response time %f s." % (time_end - time_start))
return True return True
except BaseException: except Exception as e:
logger.error("Failed to speech classification.") logger.error("Failed to speech classification.")
return False return False
@ -302,12 +285,6 @@ class CLSClientExecutor(BaseExecutor):
url = 'http://' + server_ip + ":" + str(port) + '/paddlespeech/cls' url = 'http://' + server_ip + ":" + str(port) + '/paddlespeech/cls'
audio = wav2base64(input) audio = wav2base64(input)
data = {"audio": audio, "topk": topk} data = {"audio": audio, "topk": topk}
time_start = time.time()
try: res = requests.post(url=url, data=json.dumps(data))
r = requests.post(url=url, data=json.dumps(data)) return res
# ending Timestamp
time_end = time.time()
print(r.json())
print("Response time %f s." % (time_end - time_start))
except BaseException:
print("Failed to speech classification.")

2
paddlespeech/server/engine/tts/paddleinference/tts_engine.py
Unescape View File

@ -531,4 +531,4 @@ class TTSEngine(BaseEngine):
postprocess_time)) postprocess_time))
logger.info("RTF: {}".format(rtf)) logger.info("RTF: {}".format(rtf))
return lang, target_sample_rate, wav_base64 return lang, target_sample_rate, duration, wav_base64

2
paddlespeech/server/engine/tts/python/tts_engine.py
Unescape View File

@ -250,4 +250,4 @@ class TTSEngine(BaseEngine):
logger.info("RTF: {}".format(rtf)) logger.info("RTF: {}".format(rtf))
logger.info("device: {}".format(self.device)) logger.info("device: {}".format(self.device))
return lang, target_sample_rate, wav_base64 return lang, target_sample_rate, duration, wav_base64

3
paddlespeech/server/restful/response.py
Unescape View File

@ -54,10 +54,11 @@ class ASRResponse(BaseModel):
#****************************************************************************************/ #****************************************************************************************/
class TTSResult(BaseModel): class TTSResult(BaseModel):
lang: str = "zh" lang: str = "zh"
sample_rate: int
spk_id: int = 0 spk_id: int = 0
speed: float = 1.0 speed: float = 1.0
volume: float = 1.0 volume: float = 1.0
sample_rate: int
duration: float
save_path: str = None save_path: str = None
audio: str audio: str

3
paddlespeech/server/restful/tts_api.py
Unescape View File

@ -98,7 +98,7 @@ def tts(request_body: TTSRequest):
tts_engine = engine_pool['tts'] tts_engine = engine_pool['tts']
logger.info("Get tts engine successfully.") logger.info("Get tts engine successfully.")
lang, target_sample_rate, wav_base64 = tts_engine.run( lang, target_sample_rate, duration, wav_base64 = tts_engine.run(
text, spk_id, speed, volume, sample_rate, save_path) text, spk_id, speed, volume, sample_rate, save_path)
response = { response = {
@ -113,6 +113,7 @@ def tts(request_body: TTSRequest):
"speed": speed, "speed": speed,
"volume": volume, "volume": volume,
"sample_rate": target_sample_rate, "sample_rate": target_sample_rate,
"duration": duration,
"save_path": save_path, "save_path": save_path,
"audio": wav_base64 "audio": wav_base64
} }

Write Preview
Loading…
Cancel
Save