A modern web application for transcribing audiofiles on your own server.
Powered by models like whisper-v3 and pyannote/speaker-diarization. Try it yourself!

Why Shoutout?

Shoutout is a state of the art web application for transcribing audiofiles including speaker diarization and time stamps. Due to its high accuracy level, ideal data privacy and easy navigation, shoutout is perfectly suited for transcriptions of interviews in qualitative research or sensitive corporate recordings.

Shoutout provides:

• a simple to use web-interface.
• highly accurate transcriptions leveraging the open source transcription model whisper-v3.
• automatic speaker detection including timestamps.
• perfect data privacy: Shoutout runs 100% local and does not share any of your data with external services.
• highly efficient and fast transcriptions using GPU acceleration and a scalable architecture.
• easy deployment due to a completely dockerized build.

Screenshots

Creating a new job

Web-interface with an overview of all jobs

Downloading transcripts after finishing a job

Results example:

SPEAKER_00 00:00:00
Sure, okay, so for documentation purposes. Are you okay with me recording the interview?

SPEAKER_01 00:00:12
Yes, I agree to the audio recording.

SPEAKER_00 00:00:17
Okay, then let's start. Could you first briefly introduce yourself, describe your background and what your doing...

SPEAKER_01 00:00:28
Well, my name is ...

Architecture

quickstart

Its recommended to use docker and docker-compose

Docker

To setup all services just run following command in the root directory.

Before running the following command, please update the environment variable MINIO_ENDPOINT inside the docker-compose.yml to an external reachable hostname! This container is called directly from the frontend.

If you want the worker to support your gpu you have to install the nvidia-container-toolkit on the host

docker compose -f docker-compose.prod.yml up -d

It will setup 6 containers:

1. Build of the dashboard at localhost:8000
2. PostgresDB on port 5433
3. MinIO S3 Bucket at localhost:9001
4. A MinIO client initializing the s3 bucket permissions
5. RabbitMQ at localhost:15672
6. Worker-Container (gpu support)

Development

Make sure that all services (postgres, minio, rabbitmq) are running

Dashboard

To start developing the dashboard run following commands, it will start the dev.

cd dashboard

npm i

npm run dev

Database

When making any changes to the database, be aware to migrate them!

npx prisma migrate dev --name {MigrationName}

Open http://localhost:3000

Worker

First activate and install all requirements into your virtualenv.

cd worker

pip install -r requirements.txt

To develop and test the worker just run the script without a container.

Be aware to stop the worker-container if it's running!

python3 main.py

Environment variables

Dashboard

Worker