eric/audita
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 2 Wiki Activity

Updated configuration to utilize segment ids provided in the input transcript

Browse Source
This commit is contained in:
Eric Rakestraw
2026-04-21 15:01:31 -05:00
parent 7e1a3f721a
commit ca01e46d77
13 changed files with 186 additions and 97 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
tests/test_corrections.py
View File

@@ -9,18 +9,18 @@ def _transcript():
return parse_transcript_json(
"""
[
{"speaker": "Eric", "start": 10.0, "end": 11.0, "text": "I ask Chontia for help."},
{"speaker": "Mike", "start": 0.0, "end": 1.0, "text": "Then Lyra."}
{"id": 1, "speaker": "Eric", "start": 10.0, "end": 11.0, "text": "I ask Chontia for help."},
{"id": 2, "speaker": "Mike", "start": 0.0, "end": 1.0, "text": "Then Lyra."}
]
"""
)
def test_apply_corrections_uses_threshold_and_preserves_segment_id_order():
def test_apply_corrections_uses_threshold_and_preserves_id_order():
transcript = _transcript()
corrections = [
CorrectionCandidate(
segment_id=0,
id=1,
original_text="Chontia",
corrected_text="Chauntea",
confidence=0.8,
@@ -38,7 +38,7 @@ def test_apply_corrections_ignores_below_threshold():
transcript = _transcript()
corrections = [
CorrectionCandidate(
segment_id=0,
id=1,
original_text="Chontia",
corrected_text="Chauntea",
confidence=0.79,
@@ -54,13 +54,13 @@ def test_apply_corrections_ignores_below_threshold():
def test_apply_corrections_allows_multiple_distinct_spans_in_one_segment():
transcript = _transcript()
first = CorrectionCandidate(
segment_id=0,
id=1,
original_text="Chontia",
corrected_text="Chauntea",
confidence=0.8,
)
second = CorrectionCandidate(
segment_id=0,
id=1,
original_text="help",
corrected_text="guidance",
confidence=0.9,
@@ -75,7 +75,7 @@ def test_apply_corrections_allows_multiple_distinct_spans_in_one_segment():
def test_apply_corrections_skips_missing_substring():
transcript = _transcript()
correction = CorrectionCandidate(
segment_id=0,
id=1,
original_text="Different text.",
corrected_text="Chauntea",
confidence=0.8,
@@ -85,15 +85,15 @@ def test_apply_corrections_skips_missing_substring():
assert result.transcript[0].text == "I ask Chontia for help."
assert len(result.skipped) == 1
assert result.skipped[0].segment_id == 0
assert result.skipped[0].id == 1
assert result.skipped[0].actual_text == "I ask Chontia for help."
assert "does not match any substring" in result.skipped[0].reason
def test_apply_corrections_skips_missing_segment_id():
def test_apply_corrections_skips_missing_id():
transcript = _transcript()
correction = CorrectionCandidate(
segment_id=99,
id=99,
original_text="Missing.",
corrected_text="Still missing.",
confidence=0.8,
@@ -103,14 +103,14 @@ def test_apply_corrections_skips_missing_segment_id():
assert [segment.text for segment in result.transcript] == ["I ask Chontia for help.", "Then Lyra."]
assert len(result.skipped) == 1
assert result.skipped[0].segment_id == 99
assert result.skipped[0].id == 99
assert "does not exist" in result.skipped[0].reason
def test_apply_corrections_skips_no_op():
transcript = _transcript()
correction = CorrectionCandidate(
segment_id=0,
id=1,
original_text="Chontia",
corrected_text="Chontia",
confidence=0.8,
@@ -123,16 +123,16 @@ def test_apply_corrections_skips_no_op():
assert "identical" in result.skipped[0].reason
def test_apply_corrections_skips_ambiguous_repeated_substring():
def test_apply_corrections_replaces_all_repeated_substrings():
transcript = parse_transcript_json(
"""
[
{"speaker": "Eric", "start": 0.0, "end": 1.0, "text": "Bane met Bane."}
{"id": 1, "speaker": "Eric", "start": 0.0, "end": 1.0, "text": "Bane met Bane."}
]
"""
)
correction = CorrectionCandidate(
segment_id=0,
id=1,
original_text="Bane",
corrected_text="Bain",
confidence=0.8,
@@ -140,15 +140,14 @@ def test_apply_corrections_skips_ambiguous_repeated_substring():
result = apply_corrections(transcript, [correction], confidence_threshold=0.8)
assert result.transcript[0].text == "Bane met Bane."
assert len(result.skipped) == 1
assert "multiple times" in result.skipped[0].reason
assert result.transcript[0].text == "Bain met Bain."
assert result.skipped == []
def test_apply_corrections_skips_empty_original_text():
transcript = _transcript()
correction = CorrectionCandidate(
segment_id=0,
id=1,
original_text="",
corrected_text="Chauntea",
confidence=0.8,