Feat/wfm trigger

[brosenberg42]

Issues:

• Update the WFM to support an optional TRIGGER property on any action openmpf#1574
• InProgressBatchJobsService.setProcessedAction is not called when a previous task produces no tracks. openmpf#1642
• Remove LONG_SPEAKER_ID and instead only use SPEAKER_ID openmpf#1643

Related PRs:

• Feat/wfm trigger openmpf#1644
• Feat/wfm trigger #336
• Feat/wfm trigger openmpf-contrib-components#88
• Feat/wfm trigger openmpf-cpp-component-sdk#121
• Remove TriggeredJobConfig openmpf-python-component-sdk#71
• Get rid of GetDetectionType openmpf-java-component-sdk#63
• Feat/wfm trigger openmpf-docker#197

---

This change is 

[jrobble]

Reviewed 4 of 4 files at r1, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @brosenberg42)

[jrobble]

Reviewed 50 of 50 files at r2, 2 of 2 files at r3, 2 of 2 files at r4, all commit messages.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @brosenberg42)

a discussion (no related file):
In the README for Whisper we have:

WHISPER_MODE: Determines whether Whisper will perform language detection, speech-to-text transcription, or speech translation. English-only models can only transcribe English audio. Set to LANGUAGE_DETECTION for spoken language detection, TRANSCRIPTION for speech-to-text transcription, and SPEECH_TRANSLATION for speech translation.

Update to:

WHISPER_MODE: Determines whether Whisper will perform language detection, speech-to-text transcription, or speech translation. If multiple languages are spoken in a single piece of media, language detection will detect only one of them. English-only models can only transcribe English audio. Set to LANGUAGE_DETECTION for spoken language detection, TRANSCRIPTION for speech-to-text transcription, and SPEECH_TRANSLATION for speech translation.

---

---

python/AzureSpeechDetection/README.md line 39 at r3 (raw file):

| Property Key                   | Description                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         |
|--------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| `SPEAKER_ID`              | A unique speaker identifier, of the form "`<start_offset>-<stop_offset>-<#>`, where `<start_offset>` and `<stop_offset>` are integers indicating the segment range (in frame counts for video jobs, milliseconds for audio jobs) for sub-jobs when a job has been segmented by the Workflow Manager. The final `#` portion of the ID is a 1-indexed counter for speaker identity within the indicated segment range. When jobs are not segmented, or not submitted through the Workflow Manager at all, `stop_offset` may instead be `EOF`, indicating that the job extends to the end of the file. |

Increment the outputChangedCounter due to this change. The OpenMPF major version number change will result in jobs being re-run anyway, but good to get in the habit.

[brosenberg42]

Reviewable status: 54 of 56 files reviewed, 2 unresolved discussions (waiting on @brosenberg42 and @jrobble)

a discussion (no related file):

Previously, jrobble (Jeff Robble) wrote…

In the README for Whisper we have:

WHISPER_MODE: Determines whether Whisper will perform language detection, speech-to-text transcription, or speech translation. English-only models can only transcribe English audio. Set to LANGUAGE_DETECTION for spoken language detection, TRANSCRIPTION for speech-to-text transcription, and SPEECH_TRANSLATION for speech translation.

Update to:

WHISPER_MODE: Determines whether Whisper will perform language detection, speech-to-text transcription, or speech translation. If multiple languages are spoken in a single piece of media, language detection will detect only one of them. English-only models can only transcribe English audio. Set to LANGUAGE_DETECTION for spoken language detection, TRANSCRIPTION for speech-to-text transcription, and SPEECH_TRANSLATION for speech translation.

Done.

---

---

python/AzureSpeechDetection/README.md line 39 at r3 (raw file):

Previously, jrobble (Jeff Robble) wrote…

Increment the outputChangedCounter due to this change. The OpenMPF major version number change will result in jobs being re-run anyway, but good to get in the habit.

Done.

[jrobble]

Reviewed 2 of 2 files at r5, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @brosenberg42)

[jrobble]

Reviewed 2 of 2 files at r6, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @brosenberg42)