STT Evals: Enhancements

[AkhileshNegi]

Summary

Target issue is #60

Bug Fixes

• Fixed dataset language selection to properly update all audio files in the dataset, ensuring consistency
• Improved evaluation display to gracefully handle missing ground-truth or transcription data without errors

Improvements

• Enhanced evaluation feedback system to support clearing previously set feedback status

Summary by CodeRabbit

Release Notes

• New Features

  • Added language info tooltip to the Datasets tab that displays on-demand.

• Improvements

  • Audio files automatically update to the new language when dataset language changes.
  • Evaluations tab now gracefully handles missing data with fallback displays.
  • Feedback can now be cleared or left unset for more flexible evaluation workflows.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 8fb87944-e35d-4e70-b6c5-8f233df2125e

📥 Commits

Reviewing files that changed from the base of the PR and between 3378d90 and 9d61805.

📒 Files selected for processing (1)

• app/speech-to-text/page.tsx

---

📝 Walkthrough

Walkthrough

Dataset language selection now toggles a language-info tooltip and propagates the chosen languageId to all in-memory audioFiles; evaluation feedback API/signature now accepts boolean | null and rendering logic is refactored to handle missing ground-truth or transcription fields safely.

Changes

Cohort / File(s)	Summary
Speech-to-text — Dataset UI app/speech-to-text/page.tsx	Added a togglable language-info tooltip (click-outside closes it). Changing dataset language updates languageId for all existing audioFiles in state.
Speech-to-text — Evaluations & Feedback app/speech-to-text/page.tsx	Changed updateFeedback(resultId, isCorrect, ...) to accept `isCorrect: boolean

Sequence Diagram(s)

(Skipped — changes are UI-focused and do not introduce a new multi-component sequential flow that meets diagram criteria.)

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

Possibly related PRs

• STT Evals: Enhancements #59 — Directly overlaps same file and changes (language propagation, nullable feedback, evaluations rendering); likely the closest match.
• STT Evaluation: Automated metric #54 — Related to Evaluations rendering and shared diff/segment utilities used by this change.
• STT Evals UI: Include audio play support in evaluations results #49 — Also modifies the Evaluations view and result field handling; relevant to nullable/missing-data adjustments.

Suggested labels

enhancement

Suggested reviewers

• nishika26
• Prajna1999
• vprashrex

Poem

🐇 I nudge the language, hop and bind,
Files all whisper, now aligned.
Nulls and guards keep errors at bay,
Diffs and tooltips lead the way—yay! 🎉

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.
Title check	❓ Inconclusive	The title 'STT Evals: Enhancements' is too vague and generic. While it references the Evaluations tab, it does not clearly convey the specific nature of the changes such as null handling, data consistency improvements, or the specific enhancements made.	Consider a more descriptive title like 'STT Evals: Add null handling and improve data consistency' or 'STT Evals: Handle missing ground-truth/transcription data' that better reflects the actual changes.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

• 📝 Generate docstrings (stacked PR)
• 📝 Generate docstrings (commit on current branch)

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch hotfix/stt-evals

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (1)

app/speech-to-text/page.tsx (1)

2005-2007: Remove unnecessary non-null assertion.

The non-null assertion result.is_correct! is misleading since result.is_correct can legitimately be null, and the updated updateFeedback signature now accepts boolean | null. The assertion suppresses TypeScript errors but doesn't change runtime behavior—null will still be passed.

♻️ Suggested fix

                             onBlur={(e) => {
-                               updateFeedback(result.id, result.is_correct!, e.target.value);
+                               updateFeedback(result.id, result.is_correct, e.target.value);
                             }}

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@app/speech-to-text/page.tsx` around lines 2005 - 2007, The call sites using
result.is_correct currently force a non-null assertion; remove the unnecessary
`!` so `updateFeedback(result.id, result.is_correct, e.target.value)` passes the
actual boolean | null value; update the invocation in the onBlur handler
(referencing result.is_correct and updateFeedback) to pass result.is_correct
directly and let the function's boolean | null signature handle nulls.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@app/speech-to-text/page.tsx`:
- Around line 2005-2007: The call sites using result.is_correct currently force
a non-null assertion; remove the unnecessary `!` so `updateFeedback(result.id,
result.is_correct, e.target.value)` passes the actual boolean | null value;
update the invocation in the onBlur handler (referencing result.is_correct and
updateFeedback) to pass result.is_correct directly and let the function's
boolean | null signature handle nulls.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 6f2e97f9-38f5-4226-b64f-3f7a4ad6ac44

📥 Commits

Reviewing files that changed from the base of the PR and between a0ada14 and 3378d90.

📒 Files selected for processing (1)

• app/speech-to-text/page.tsx

[nishika26]

Maybe adding an info tip on language field of dataset creation would be helpful where we tell them this is the default language that will be considered for all smaples if they have not explicitely stated sample languages

[AkhileshNegi]

sure will add