susom/docker-ffmpeg

Name: docker-ffmpeg

Owner: Stanford School of Medicine

Description: a ffmpeg web service

Created: 2017-05-17 18:39:22.0

Updated: 2018-04-25 22:27:38.0

Pushed: 2017-05-17 18:56:35.0

Homepage: null

Size: 6

Language: PHP

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

docker-ffmpeg

a ffmpeg web service

This is a container that does two things:

1) It wraps ffmpeg as a web service so you can post audio and get back a compressed version of the audio in a variety of formats 2) It integrates with IBM Bluemix Watson to provide audio to text capability

https://www.ibm.com/watson/developercloud/doc/speech-to-text/index.html

Building Container

er build -t ffmpeg-web-service:latest .

Running Container

er run -d --name 'ffmpeg' \
bluemix_user_pass=xxx:yyy" \
080:80 \
start 'unless-stopped' \
eg-web-service:latest

where xxx:yyy equals the service credentials provided from your watson service. You must first create a watson service with the speech-to-text engine. From here you should be able to provision credentials like the following:

rl": "https://stream.watsonplatform.net/speech-to-text/api",
sername": "xxx",
assword": "yyy"

Web Service

This web service runs FFMPEG to convert a single input file into a single output file.

:   The upload file contents with a valid file suffix to indicate the format
on: TRANSCODE (default) or TRANSCRIBE

Transcoding:
format: The output file format (wav/mp3/etc...)
rate:   (optional) override the encoding rate in Hz (e.g. 16000)

Transcribing:
language: en (default), es (spanish), zh (chinese), pt (portuguese)

Example Request:

POST:

    file => "audio_1.wav"
    format => 'mp3'
    rate => '48000'

Will return a streamed MP3 file or error message

POST:

    file => "audio_1.amr"
    language => 'es'

Will return a json object for the translation of the audio, such as:

"results": [
    {
        "alternatives": [
            {
                "confidence": 0.28,
                "transcript": "flat out a few birds "
            }
        ],
        "final": true
    }
],
"result_index": 0

Since the transcription service only accepts 16kHz or higher samples, in the example above, the 8 Khz amr file will be transiently transcoded to a 16kHz wav file and then sent off for transcription. This can take 30 seconds or longer.