Skip to content

[BUG] Transcription prompt hardcodes "Jasper" — should use ASSISTANT_NAME #5

Closed

Description

@Harshbansal8705

Description

config.py L84-88 has the Whisper transcription prompt hardcoded with "Jasper":

TRANSCRIPTION_PROMPT = (
    "Please transcribe the following audio accurately, maintaining proper "
    "punctuation and formatting. This is a conversation between a user and "
    "a Desktop Assistant named \"Jasper\". So, focus on words like 'Jasper'."
)

Steps to Reproduce

  1. Set ASSISTANT_NAME="Atlas" in .env
  2. Speak to the assistant using the custom wake word
  3. Observe that the transcription prompt still tells Whisper to focus on "Jasper"

Expected Behavior

The transcription prompt should dynamically use ASSISTANT_NAME so that users with custom wake words get accurate transcription.

Suggested Fix

TRANSCRIPTION_PROMPT = (
    "Please transcribe the following audio accurately, maintaining proper "
    "punctuation and formatting. This is a conversation between a user and "
    f"a Desktop Assistant named \"{ASSISTANT_NAME}\". So, focus on words like '{ASSISTANT_NAME}'."
)

Note: Since ASSISTANT_NAME is defined as a class attribute above this line, it can reference it directly.

Files to Modify

  • src/config.py

Activity

  1. sivaadithya25 commented on Sep 25, 2026

    @sivaadithya25
    Contributor

    Hi @Harshbansal8705, I'd like to work on this issue. I'll replace the hardcoded "Jasper" reference with the configurable ASSISTANT_NAME value to keep the transcription prompt consistent with the project configuration. Could you please assign this issue to me?

  2. Harshbansal8705 commented on Sep 25, 2026

    @Harshbansal8705
    OwnerAuthor

    Sure. I think you can also resolve #3 in the same PR.

  3. sivaadithya25 commented on Sep 25, 2026

    @sivaadithya25
    Contributor

    Description

    Fixes #3
    Fixes #5

    Replaced hardcoded references to "Jasper" with config.ASSISTANT_NAME to ensure personalization settings are respected throughout the application.

    Type of Change

    • Bug fix (non-breaking change that fixes an issue)

    Changes Made

    Issue #3 - Overlay UI

    • Updated overlay window title to use config.ASSISTANT_NAME
    • Updated response message prefix to use config.ASSISTANT_NAME
    • Updated active status text to use config.ASSISTANT_NAME

    Issue #5 - Transcription Prompt

    • Updated transcription prompt to use config.ASSISTANT_NAME
    • Removed hardcoded assistant name from Whisper transcription instructions

    How Has This Been Tested?

    • Tested the specific feature/fix manually
    • Verified no regressions in existing functionality

    Validation:

    python -m py_compile src/config.py
    python -m py_compile src/ui/overlay.py
  4. sivaadithya25 commented on Oct 5, 2026

    @sivaadithya25
    Contributor

    Hi @Harshbansal8705,

    Hope you're doing well! Now that October has started and OSCI'26 is resuming, I wanted to kindly follow up on PR (#14) when you have a moment.

    I noticed there are some merge conflicts in src/ui/overlay.py and a failing lint check. Since you mentioned you were working on updating the setup and linting of the project, please let me know if I should go ahead and resolve the conflict/lint issues from my end, or if I should wait for your updates.

    Thank you! 😊

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

OSCI'26Open Source Contribution Initiative 2026bugSomething isn't workinggood first issueGood for newcomers

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions