Using Privox

Privox is accessed by doing an HTTP post to one of the public APIs (STT or TTS) and providing the proper input for the operation. These are summarized below.

STT

key

The user’s API key. This key may be found on your profile page.

default = none, this is a required parameter

quality

One of ‘fast’, ‘normal, ‘better’, best’, ‘xcribe’ or ‘xcribe2’ representing the quality of transcription being requested from least to most accurate.

default = fast

language

The language of the resultant text being requested. If you specifiy a different language than what you provided in the wav file, the system will attempt to translate. This does not work well with the smaller/faster models. Transcription will be faster for english if you specify english.

default = English

xlate

If this field is posted it must contain the value ‘xlate’. It is used to tell the API the wav data is not in the proper format and it must be converted before being interpreted. This adversely impacts performance and should be avoided. It is typically used by Javascript.

default = None

Examples

Curl

# minimum input
curl -X POST http://api.privox.io/stt -F "file=@wav_files/sample.wav" -F "key=X"

# normal English usage
curl -X POST http://api.privox.io/stt -F "file=@wav_files/sample.wav" -F "quality=fast" -F "language=English" -F "key=X"

# normal French usage
curl -X POST http://api.privox.io/stt -F "file=@wav_files/parlez_vous.wav" -F "quality=best" -F "language=French" -F "key=X"

# transcription quality
curl -X POST http://api.privox.io/stt -F "file=@wav_files/sample.wav" -F "quality=xcribe2" -F "language=English" -F "key=X"

Bash

# test and time using compression
# usage ./this_file.sh API_KEY FILENAME.WAV
date
cp $2 sample.wav
zip archive.zip sample.wav
curl -X POST -F "file=@archive.zip" -F "quality=normal" -F "language=en" -F "key=$1" api.privox.io/stt
date

Python

import requests
# stt example

stt_uri = 'http://api.privox.io/cgi-bin/stt.py'

with open('mydata.wav')as f: data = f.read()

form = {
  'file': data,
  'key': 'my-api-key',
  'model': 'fast'
  }

response = requests.post(stt_uri, files=form)

print(response.content.decode('utf-8').strip())

TTS

key

The user’s API key. This key may be found on your profile page.

default = none, this is a required parameter

voice

Either ‘voice1’ or ‘voice2’.

default = voice1

language

The language for the voice. Currently ignored.

default = English

text

The text string to convert to a wav file.

default = none, this is a required parameter

Examples

Curl

# minimum input
curl -X POST http://api.privox.io/tts -H "Content-Type: application/x-www-form-urlencoded" -d "key=XXX&text=Test utterance." | aplay

# normal usage
curl -X POST http://api.privox.io/tts -H "Content-Type: application/x-www-form-urlencoded" -d "key=XXX&language=en&voice=voice2&text=Test utterance." > junk.wav

# time it
date;curl -X POST http://api.privox.io/tts -H "Content-Type: application/x-www-form-urlencoded" -d "key=XXX&text=Test utterance." > tts.wav;date;aplay tts.wav