VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages.
Python aiaudiobookcudadubbingelevenlabs-alternativehuggingfacelocal-firstmlxomnivoice-studiospeech-to-texttext-to-speechtranscriptiontranslatettsvoice-aivoice-cloningvoice-generationvoicestudioworkflow
"""Cleanup must retain attributed non-ultra-short speaker turns."""
import pytest
@pytest.mark.parametrize("answer_first", [False, True])
def test_cleanup_keeps_short_answer_attributed_to_its_speaker(answer_first):
from services.segmentation import clean_up_segments
whole = {"id": "whole", "start": 0, "end": 3, "text": "First speaker delivers a whole sentence.", "speaker_id": "Speaker 1"}
answer = {"id": "answer", "start": 3, "end": 4, "text": "Second speaker answers right away", "speaker_id": "Speaker 2", "translations": {"fr": "Une autre voix"}}
if answer_first:
answer.update(start=0, end=1)
whole.update(start=1, end=4)
source = [answer, whole] if answer_first else [whole, answer]
assert clean_up_segments(source) == source
def test_same_speaker_short_answer_still_merges():
from services.segmentation import clean_up_segments
source = [{"start": 0, "end": 3, "text": "First speaker delivers a whole sentence.", "speaker_id": "Speaker 1"}, {"start": 3, "end": 4, "text": "Second clause follows right away", "speaker_id": "Speaker 1"}]
out = clean_up_segments(source)
assert len(out) == 1
assert out[0]["text"] == " ".join(s["text"] for s in source)
def test_documented_ultra_short_stray_token_still_folds():
from services.segmentation import clean_up_segments
source = [{"start": 0, "end": 3, "text": "First speaker delivers a whole sentence.", "speaker_id": "Speaker 1"}, {"start": 3, "end": 3.2, "text": "STR", "speaker_id": "Speaker 2"}]
out = clean_up_segments(source)
assert len(out) == 1
assert out[0]["text"].endswith(" STR")