TRANSCRIBER TO BITCOIN TRANSCRIPT

This cli app transcribes audio and video for submission to the bitcointranscripts repo.

Available transcription models and services

• (local) Whisper --model xxx [default: tiny.en]
• (remote) Deepgram (whisper-large) --deepgram [default: False]
  • summarization --summarize
  • diarization --diarize

Transcription Workflow

This transcription tool operates through a structured four-stage process:

1. Preprocess: Gathers all the available metadata for each source (supports YouTube videos&playlists, and RSS feeds)
2. Process: Downloads and converts sources for transcription preparation
3. Transcription: Utilizes openai-whisper or Deepgram to generate transcripts.
   1. Converts audio to text.
      • Save as JSON: Preserves the output of the transcription service for future use.
      • Save as SRT: Generates SRT file [whisper only]
   2. Summarize: Generates a summary of the transcript. [deepgram only]
   3. Upload: Saves transcription service output in an AWS S3 Bucket [optional]
   4. Finalizes the resulting transcript.
      • Process diarization. [deepgram only]
      • Process chapters.
4. Postprocess: Offers multiple options for further actions:
   • Push to GitHub: Push transcripts to your fork of the bitcointranscripts repo.
   • Markdown: Saves transcripts in a markdown format supported by bitcointranscripts.
   • Upload: Saves transcripts in an AWS S3 Bucket.
   • Push to Queuer backend: Sends transcripts to a Queuer backend.
   • Save as JSON: Preserves transcripts for future use.

Prerequisites

• To use deepgram as a transcription service, you must have a valid DEEPGRAM_API_KEY in the .env file.

• To push the resulting transcript to GitHub you need to fork bitcointranscripts and then clone your fork and define your BITCOINTRANSCRIPTS_DIR in the .env file.

• To push the resulting transcript to a Queuer backend, you must have a valid QUEUE_ENDPOINT in the .env file. If not, you can instead save the payload in a json file using the --noqueue flag.

• To enable pushing the models to a S3 bucket,

  • Install aws-cli to your system.
  • Configure aws-cli by first generating IAM credentials (if not already present) and using aws configure to set them.
  • To verify proper configuration, run aws s3 ls to show the list of S3 buckets. Don't forget to set a valid S3_BUCKET in the .env file.

• To be able to convert the intermediary media files to mp3, install FFmpeg

  • for Mac Os users, run brew install ffmpeg

  • for other users, follow the instruction on their site to install

Install/Uninstall

Navigate to the application directory and run the below commands:

python3 -m venv venv creates a virtual environment

source venv/bin/activate activates the virtual environment

pip3 install . --use-pep517 to install the application

To check the version: tstbtc --version view the application version

tstbtc --help view the application help

pip3 uninstall tstbtc to uninstall the application

Usage

tstbtc transcribe {source_file/url} transcribe the given source

Suported sources:

• YouTube videos and playlists
• Local and remote audio files
• JSON files containing individual sources

Note:

• The https links need to be wrapped in quotes when running the command on zsh

To include optional metadata in your transcript, you can add the following parameters:

• --loc: Add the location in the bitcointranscripts hierarchy that you want to associate the transcript [default: "misc"]
• -t or --title: Add the title for the resulting transcript (required for audio files)
• -d or --date: Add the event date to transcript's metadata in format 'yyyy-mm-dd'
• can be used multiple times:
  • -T or --tags: Add a tag to transcript's metadata
  • -s or --speakers: Add a speaker to the transcript's metadata
  • -c or --category: Add a category to the transcript's metadata

To configure the transcription process, you can use the following flags:

• -m or --model: Select which whisper model to use for the transcription [default: tiny.en]
• -D or --deepgram: Use deepgram for transcription, instead of using the whisper model [default: False]
• -M or --diarize: Supply this flag if you have multiple speakers AKA want to diarize the content [only available with deepgram]
• -S or --summarize: Summarize the transcript [only available with deepgram]
• --github: Specify the GitHub operation mode
• -u or --upload: Upload processed model files to AWS S3
• --markdown: Save the resulting transcript to a markdown format supported by bitcointranscripts
• --noqueue: Do not push the resulting transcript to the Queuer, instead store the payload in a json file
• --nocleanup: Do not remove temp files on exit

Examples

To transcribe this podcast episode from YouTube from Stephan Livera's podcast and add the associated metadata, we would run either of the below commands. The first uses short argument tags, while the second uses long argument tags. The result is the same.

• tstbtc transcribe Nq6WxJ0PgJ4 --loc "stephan-livera-podcast" -t 'OP_Vault - A New Way to HODL?' -d '2023-01-30' -T 'script' -T 'op_vault' -s 'James O’Beirne' -s 'Stephan Livera' -c ‘podcast’
• tstbtc transcribe Nq6WxJ0PgJ4 --loc "stephan-livera-podcast" --title 'OP_Vault - A New Way to HODL?' --date '2023-01-30' --tags 'script' --tags 'op_vault' --speakers 'James O’Beirne' --speakers 'Stephan Livera' --category ‘podcast’

You can also transcribe a remote audio/mp3 link, such as the following from Stephan Livera's podcast:

mp3_link="https://anchor.fm/s/7d083a4/podcast/play/64348045/https%3A%2F%2Fd3ctxlq1ktw2nl.cloudfront.net%2Fstaging%2F2023-1-1%2Ff7fafb12-9441-7d85-d557-e9e5d18ab788.mp3"
tstbtc transcribe $mp3_link --loc "stephan-livera-podcast" --title 'SLP455 Anant Tapadia - Single Sig or Multi Sig?' --date '2023-02-01' --tags 'multisig' --speakers 'Anant Tapadia' --speakers 'Stephan Livera' --category 'podcast'

Testing

To run the unit tests

pytest -v -m main -s

To run the feature tests

pytest -v -m feature -s

To run the full test suite

pytest -v -s

License

Transcriber to Bitcoin Transcript is released under the terms of the MIT license. See LICENSE for more information or see https://opensource.org/licenses/MIT.