SDK

Languages/Frameworks	One-sentence Recognition Real-time Speech Transcription	Audio File Transcription
Vue	To be done	N/A
JavaScript	To be done	N/A
Android	To be done	N/A
C++	To be done	N/A
Java	Download	N/A
Python	Download	Example code can be found below

Note:

In the source code, Real-Time Speech Transcription is represented by Transcriber, and One-sentence Recognition is represented by Recognizer. Please refer to these as needed.

The Audio File Transcription service is provided as a standard HTTP interface, so no SDK is offered. You can develop the integration based on the interface protocol.

Audio File Transcription (Form-data) Example Codes for Python

#!/usr/bin/env python
# -*- encoding: utf-8 -*-
import time
import requests
from urllib3 import encode_multipart_formdata
host = 'http://127.0.0.1' #Server IP
port = 7100
#Set parameters
params = {
    "file": ("filename", open('d:/test.wav', 'rb').read()),
    'lang_type': 'ja-JP',
    'sample_rate': 16000,
    'format': 'wav',
    'enable_inverse_text_normalization': True,
    'paragraph_condition': 300,    #Paragraph characters count
    ########## Configure as following for a far-field recording
    #'max_sentence_silence': 3000,
    #'gain': 5,
    #'decode_silence_segment': True
}
encode_data = encode_multipart_formdata(params)
data = encode_data[0]
headers = {'Content-Type': encode_data[1], 'Request-Index': '0'}
request_url = f'{host}:{port}/request'
#Create an Audio File Transcription task 
response = requests.post(request_url, data=data, headers=headers)
task_id = response.json()['data']['task_id']
get_result_url = f'{host}:{port}/getResult?task_id={task_id}'
start_time = time.time()
#Poll to get transcription results
while True:
    formatted_time = time.strftime('%H:%M:%S', time.gmtime(time.time() - start_time))
    data = requests.get(get_result_url).json()
    if data['status'] == '00000':
        print(f'\r{formatted_time} Processing... 100%', end='')
        break
    elif data['status'] == '20302':
        progress = data['data']['progress']
        #Processing
        if progress > 0:
            print(f'\r{formatted_time} Processing... {progress}%' + ' ' * 25, end='')
        #Queuing
        elif progress < 0:
            print(f'\r{formatted_time} Waiting in queqe, with {-progress} task(s) ahead.', end='')
        #Processing failed
        else:
            print('\n',data)
            break
        #Query the results again after 1 second
        time.sleep(1)
    else:
        print('\n',data)
        break
data_dict = dict()
print('\n')
#Print the result
for i in data['data']:
    paragraph = i.get("paragraph", 0)
    data_dict[paragraph] = data_dict.get(paragraph, '') + ' ' + i['transcript']
for seg_index, result in data_dict.items():
    print(f'\t{result}')

Audio File Transcription (Form-data) Example Codes for Python

目次