omakase-media-tools 0.1.16

A CLI utility and samples for creation of sample media and temporal metadata tracks for use with the Omakase Player framework

Omakase Media Tools Utility

This PIP package contains the omt utility and sample media reference for use with the Omakase Player and Omakase Reference Player open source projects.

The sample media is provided to demonstrate the capabilities of the Omakase Player framework and to bootstrap small POC projects. The sample media reference and omt utility are provided as-is and are not intended for use in production environments.

The Python omt.py utility and documentation can be used to create your own sample media, temporal metadata tracks and OMP player JSON for use with the Omakase Player framework and the Omakase Reference Player.

The utility can create an OMP player json file from a template, source media and HLS media:

• OMP Player JSON using the omt player-json CLI command

The utility can generate the following types of metadata tracks:

• Audio Metric Analysis Tracks using the omt audio-metrics CLI command
• Audio Waveform Analysis Tracks using the omt waveforms CLI command
• Video Bitrate Analysis Tracks using the omt video-bitrate CLI command
• Video Thumbnail Tracks using the omt thumbnails CLI command

The reference implementation of the Omakase player can be found in GitHub at omakase-reference-player.

Contents:

• Requirements
• Installation
• Usage
  • Usage omt player-json
  • Usage omt thumbnails
  • Usage omt video-bitrate
  • Usage omt audio-metrics
  • Usage omt waveforms
• External Links
• License

Requirements

---

Python

• Python 3.8 or higher

ffmpeg

• Download a static build from the ffmpeg website and install using the instructions for your platform.
• Ensure that the ffmpeg executable is in your path.

audiowaveform

audiowaveform is a C++ program that takes an audio file and generates raw waveform data from it.

This data is then used to generate an OMP v1.0 VTT file containing the waveform metadata needed by the Omakase Player.

• download and install from GitHub here: audiowaveform

ffmpeg-bitrate-stats

The ffmpeg-bitrate-stats Python package is used to generate raw video bitrate metrics, when are then processed to create an OMP v1.0 VTT file for use with Omakase Player.

• Install the ffmpeg-bitrate-stats Python package from the following GitHub repository: ffmpeg-bitrate-stats

MediaInfo

The mediainfo CLI utility is used to extract media file information, that is then used to populate metadata in the OMP player json file.

IMPORTANT: The CLI version of MediaInfo must be installed on your system and available in your path. This installation is in addition to the GUI version of MediaInfo that may already be installed on your system.

• Download and install the CLI VERSION from MediaInfo website for your platform.

Installation

---

• Install the PIP package omp-media-tools from PyPi.

Usage

---

The Python utility omt.py is a command line utility that can generate Omakase Player temporal metadata tracks and OMP player json files from source media.

The utility can create an OMP player json file from a template, source media and HLS media:

• OMP Player JSON using the omt player-json CLI command

The utility can generate the following types of metadata tracks:

• Audio Metric Analysis Tracks using the omt audio-metrics CLI command
• Audio Waveform Analysis Tracks using the omt waveforms CLI command
• Video Bitrate Analysis Tracks using the omt video-bitrate CLI command
• Video Thumbnail Tracks using the omt thumbnails CLI command

Usage omt player-json

To assist with creation of an OMP player json file, the omt player-json utility command can be used to generate the OMP player json file from a template.

The template file contains the minimal essential information required to generate the OMP player json. With a default directory structure to hold the source media, HLS media and metadata tracks, the omt player-json utility command can search the directories for the required file references, create required technical metadata and generate the OMP player json.

Default Directory Structure

The omt player-json utility command expects the default directory structure as shown below. The OMP player json does not require this structure, but using this default makes a sparse template json possible.

The utility command should be run from the working_directory directory where the template.json file is located. This is passed as follows:

• omt player-json --template template.json

The media_root directory is the root directory for all HLS media, analysis metadata tracks, thumbnails and waveforms. This directory will be uploaded to the CDN or webserver.

The analysis directory contains the audio and video analysis metadata tracks. This includes metadata tracks created with the omt utility as well as those created manually or through some other process.

The metadata tracks created with the following omt utility commands should be placed in the analysis directory:

• omt audio-metrics for the EBU R128 and RMS Levels audio metric metadata tracks
• omt video-bitrate for the video bitrate metadata track

The thumbnails directory contains the video thumbnail metadata track and images created with the omt thumbnails utility command.

The waveforms directory contains the audio waveform metadata track created with the omt waveforms utility command.

The hls directory contains a directory for each HLS ABR rendition to be included. During creation of the OMP player json, the hls directories are searched for media files to be referenced.

The sources directory contains the source media files to be used to generate the HLS ABR renditions. These files are referenced to create audio and video metadata files and as a source of technical metadata (MediaInfo) to be presented within the player.

working_directory/
    ├── root_dir/
    │   └── player.json
    │   └── analysis/
    │   └── thumbnails/
    │   └── waveforms/
    │   └── hls/
    │       └── hls_abr_1/
    │       └── hls_abr_2/
    ├── sources/
    └── template.json

Template Structure and Example

An example template.json file is shown below. The template.json file contains the minimal essential information needed to create an OMP player json.

The output section contains several key fields used by the omt player-json utility command to generate the OMP player json and navigate the directory structure.

The root_dir field specifies the name of the root directory (see above). The sources_dir field specifies the name where the mezzanine files are located (see above).

The root_url field specifies the URL root that prepended to all file references in the OMP player json. For example, it the root_url is https://localhost:8080/, the url reference to the HLS m3u8 manifest tears-of-steel.m3u8 for the tears-of-steel_sdr_1080p24_BITC rendition would be:

• https://localhost:8080/hls/tears-of-steel_sdr_1080p24_BITC/tears-of-steel.m3u8.

The mezzanine section contains the names of the source media files and assigned an ID to them. The omt player-json utility command will search the sources directory for these files. Video files are assigned an ID starting with V, audio files are assigned an ID starting with A and text files are assigned an ID starting with T.

The hls section contains the directories holding the HLS ABR renditions to be included in the player. Each rendition is assigned an ID and a display name for presentation in the Omakase Player. The omt player-json utility command will search the hls directory for these directories.

The tracks section contains the video, audio and text tracks to be presented in OMP player. Each track is mapped to the source mezzanine file via the source_id field and the media track in the HLS ABR m3u8 manifest via the program_name field. Thedisplay_text field is used to present the track name in OMP player.

{
    "sources": {
        "mezzanine": [
            {
                "id": "V1",
                "src": "tearsofsteel_4k.mov"
            },
            {
                "id": "A1",
                "src": "Surround-TOS_DVDSURROUND-Dolby%205.1.ac3"
            },
            {
                "id": "T1",
                "src": "TOS-en.srt"
            }
        ],
        "hls": [
            {
                "id": "HLS-1080",
                "display_name": "Confidence QC 1080p",
                "src": "tears-of-steel_sdr_1080p24_BITC"
            },
            {
                "id": "HLS-720",
                "display_name": "Proxy 720p",
                "src": "tears-of-steel_sdr_720p24_BITC"
            }
        ],
        "tracks": {
            "video": [
                {
                    "source_id": "V1"
                }
            ],
            "audio": [
                {
                    "source_id": "V1",
                    "display_text": "English 2.0",
                    "program_name": "EN_20"
                },
                {
                    "source_id": "A1",
                    "display_text": "English 5.1",
                    "program_name": "EN_51"
                }
            ],
            "text": [
                {
                    "source_id": "T1",
                    "display_text": "English Subtitles",
                    "program_name": "English Subtitles"
                }
            ]
        },
        "metadata": [
            {
                "display_name": "OMDb Title Metadata",
                "src": "omdb.json"
            }
        ]
    },
    "output": {
        "root_dir": "tears-of-steel",
        "sources_dir": "sources",
        "root_url": "https://localhost:8080/"
    }
}

Usage omt thumbnails

A thumbnail timeline metadata track and series of thumbnail images, can be generated with this command.

An input video file must be specified and the VTT and image files generated are written to the output directory specified.

The frequency of the thumbnails is controlled by the --seconds-interval CLI option.

usage: omt.py thumbnails [-h] [-v] -i INPUT -o OUTPUT
                         [-s {1,2,3,4,5,10,12,15}]

options:
  -h, --help            show this help message and exit
  -v, --verbose         enable verbose output
  -i INPUT, --input INPUT
                        input video file
  -o OUTPUT, --output OUTPUT
                        output directory
  -s {1,2,3,4,5,10,12,15}, --seconds-interval {1,2,3,4,5,10,12,15}
                        seconds between video thumbnails

Example - Thumbnail Timeline Metadata Track

WEBVTT

00:00:00.000 --> 00:00:01.999
tearsofsteel_4k00001.jpg

00:00:02.000 --> 00:00:03.999
tearsofsteel_4k00002.jpg

Usage omt video-bitrate

A metadata track of the video bitrate can be generated with this utility command. The resulting metadata track can then be used to visualize the video bitrate as a line chart on the Omakase Player timeline.

usage: omt.py video-bitrate [-h] [-v] -i INPUT -o OUTPUT
                            [-s {1,2,3,4,5,10,12,15}]

options:
  -h, --help            show this help message and exit
  -v, --verbose         enable verbose output
  -i INPUT, --input INPUT
                        input video file
  -o OUTPUT, --output OUTPUT
                        output directory
  -s {1,2,3,4,5,10,12,15}, --seconds-interval {1,2,3,4,5,10,12,15}
                        seconds between bitrate samples

An input video file must be specified and the VTT and image files generated are written to the output directory specified.

The resolution of the bitrate samples is controlled by the --seconds-interval CLI option.

Example - Video Bitrate Metadata Track

WEBVTT

NOTE
Omakase Player Web VTT
V1.0

00:00:00.000 --> 00:00:01.999
276.39:MEASUREMENT=avg:COMMENT=2-sec interval

Please Note: The OMP v1.0 VTT file format is a standard WebVTT file with the following additional metadata:

• The :MEASUREMENT=<metric name> tag is optional and can be used to specify the video bitrate metric.
• The :COMMENT=<comment> tag is optional indicates the sample interval for the bitrate metric.

The optional tags are used by the Omakase Player to provide telemetry metadata for the video bitrate metric as the video is played.

Usage omt audio-metrics

Audio metrics can be generated for an audio tracks, or all audio tracks, with this CLI command. At present, two audio metric metadata tracks are created for each audio file:

• RMS Levels using the ffmpeg ametadata filter
• R128 Momentary Loudness also using the ffmpeg ametadata filter

If --input is a directory, all of the wav and aac files in the current directory are processed. If --input is a file, only the audio file specified is processed.

The resulting metadata tracks are named with the basename of the audio file and appended with R128_2-SEC or RMS_2-SEC respectively. All files are written to the directory specified with --output.

At present, the metrics are calculated as an average over a two-second interval.

usage: omt.py audio-metrics [-h] [-v] -i INPUT -o OUTPUT

options:
  -h, --help            show this help message and exit
  -v, --verbose         enable verbose output
  -i INPUT, --input INPUT
                        input media file or directory
  -o OUTPUT, --output OUTPUT
                        output directory

Example - Audio Metrics Metadata Track

WEBVTT

NOTE
Omakase Player Web VTT
V1.0

00:00:00.000 --> 00:00:01.999
-56.033:MEASUREMENT=lavfi.r128.M:COMMENT=2-sec avg

Please Note: The OMP v1.0 VTT file format is a standard WebVTT file with the following additional metadata:

• The :MEASUREMENT=<metric name> tag is optional and can be used to specify the audio metric type.
• The :COMMENT=<comment> tag is optional and can be used to provide additional information about the audio metric.

The optional tags are used by the Omakase Player to provide telemetry metadata for the audio metric as the video is played.

Usage omt waveforms

The waveforms command generates audio waveform metadata tracks that can be used to provide an audio waveform visualization in Omakase Player.

Waveform metadata can be generated for a single audio track, or all audio tracks, with this omt command. The generated waveform includes the entire soundfield, but individual channels can be visualized with a dual-mono audio track for each channel.

If --input is a directory, all of the wav and aac files in the current directory are processed. If --input is a file, only the audio file specified is processed.

The resulting metadata tracks are named with the basename of the audio file All files are written to the directory specified with --output.

At present, the metrics are calculated as an average over a 1-second interval.

usage: omt.py waveforms [-h] [-v] -i INPUT -o OUTPUT

options:
  -h, --help            show this help message and exit
  -v, --verbose         enable verbose output
  -i INPUT, --input INPUT
                        input media file or directory
  -o OUTPUT, --output OUTPUT
                        output directory

Example - Audio Waveform Metadata Track

WEBVTT
  
00:00:00.000 --> 00:00:00.999
-0.0101, 0.0108

External Links

---

• Omakase Player Project Page
• Omakase Player GitHub Repository
• Omakase Reference Player GitHub Repository
• ffmpeg Project Page
• ffmpeg-bitrate-stats GitHub Repository
• audiowaveform Project Page
• audiowaveform GitHub Repository## License

License

---

omakase-media-tools, Copyright 2024 ByOmakase, LLC (https://byomakase.org)