Python Script for Video and Audio transcription
Budget: $10 – $100 USD
I'm looking for a seasoned Python developer with specific experience in video analysis to create a script that can transcribe a full mp4 video and audio into text.
Key Requirements:
- The script should be able to read a video in mp4, mkv, mov format or a audio (AAC, mp3, ...)
- The library used for reading the video is not fixed. You can choose from libraries which should be compatible with both Windows and Linux (and mac)
- take a look here: https://github.com/kkroening/ffmpeg-python/blob/master/examples/transcribe.py
- The output should detail the (as a class, TranscriptionDetails)
... - file path
... - file name
... - frame number / or location in the video/audio
... - time code / or location in the video/audio
... - the transcription
... - confidence of transcription
and should be outputed in a JSON file.
The vidoes may have no (0) voices or mutiple person voices (n)
Ideal Skillset:
- Proficiency in Python
- Experience with audio processing libraries
- Experience with transcription services processing libraries
- Experiences with OOP (classes and lists)
- Familiarity with JSON output format.
Milestones:
MS1:
extract transcription of the voices (0..n voices)
we will share you testing videos and audios
MS2:
create a new video based on the json and add the transcription as a overlay to the video
Budget?
will not be disclosed, place your best bid to get considered
Payment?
- you estimate in a WBS (optimistic, expected, pessimistic, where optimistic < expected < pessimistic) after getting the task
- we discuss about clearances and effort
- we mutually agree to effort
- we assign you the task after mutually agreed
- you provide us a delivery date
- you implement, test and deliver
- you provide also unit /integration tests to ensure code is always working
- you document each required step to make interaction of systems to work
- we test on our premise (you maybe help to fix issues)
- we pay
(basically the rules of freelancer)
Closed book vs open book?
We work only on open book.
Closed book means you are unwilling to define a WBS for the work and you add only a price tag to the task.
We are sorry we will not hire you in such a case!
Deliveries?
- in our on premise git (access will be granted to you)
- full sources (NO! Jupyter Notebook)
- multiple samples of videos will be shared via URL to download
- bash to build & run your code
- libs, need prior confirm and we prefer to use mostly latest stable versions
- your code is working on our end too
- your code passes our CI build (with provided maven pom.xml can be checked)
- you may require to help us here on setup
- we require, depending on availability and on complexity 2-4 weeks test time per milestone. Usually clearly less
Key Requirements:
- The script should be able to read a video in mp4, mkv, mov format or a audio (AAC, mp3, ...)
- The library used for reading the video is not fixed. You can choose from libraries which should be compatible with both Windows and Linux (and mac)
- take a look here: https://github.com/kkroening/ffmpeg-python/blob/master/examples/transcribe.py
- The output should detail the (as a class, TranscriptionDetails)
... - file path
... - file name
... - frame number / or location in the video/audio
... - time code / or location in the video/audio
... - the transcription
... - confidence of transcription
and should be outputed in a JSON file.
The vidoes may have no (0) voices or mutiple person voices (n)
Ideal Skillset:
- Proficiency in Python
- Experience with audio processing libraries
- Experience with transcription services processing libraries
- Experiences with OOP (classes and lists)
- Familiarity with JSON output format.
Milestones:
MS1:
extract transcription of the voices (0..n voices)
we will share you testing videos and audios
MS2:
create a new video based on the json and add the transcription as a overlay to the video
Budget?
will not be disclosed, place your best bid to get considered
Payment?
- you estimate in a WBS (optimistic, expected, pessimistic, where optimistic < expected < pessimistic) after getting the task
- we discuss about clearances and effort
- we mutually agree to effort
- we assign you the task after mutually agreed
- you provide us a delivery date
- you implement, test and deliver
- you provide also unit /integration tests to ensure code is always working
- you document each required step to make interaction of systems to work
- we test on our premise (you maybe help to fix issues)
- we pay
(basically the rules of freelancer)
Closed book vs open book?
We work only on open book.
Closed book means you are unwilling to define a WBS for the work and you add only a price tag to the task.
We are sorry we will not hire you in such a case!
Deliveries?
- in our on premise git (access will be granted to you)
- full sources (NO! Jupyter Notebook)
- multiple samples of videos will be shared via URL to download
- bash to build & run your code
- libs, need prior confirm and we prefer to use mostly latest stable versions
- your code is working on our end too
- your code passes our CI build (with provided maven pom.xml can be checked)
- you may require to help us here on setup
- we require, depending on availability and on complexity 2-4 weeks test time per milestone. Usually clearly less