From eb08c1fef950d5e1cb754167af6db795b66caa44 Mon Sep 17 00:00:00 2001
From: itqop <leonkl32@gmail.com>
Date: Thu, 19 Feb 2026 04:11:36 +0300
Subject: [PATCH] fixes bugs

---
 README.md                                  | 5 ++---
 requirements.txt                           | 1 -
 transcriber/diarization/pyannote_engine.py | 4 ++--
 3 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/README.md b/README.md
index 3efad96..811f89a 100644
--- a/README.md
+++ b/README.md
@@ -5,7 +5,7 @@
 ## Возможности
 
 - Распознавание речи (faster-whisper, модель large-v3)
-- Диаризация спикеров (pyannote.audio 3.1)
+- Диаризация спикеров (pyannote.audio community-1)
 - Поддержка форматов: `.m4a`, `.mp3`, `.wav`, `.aac`
 - Аудио до 3 часов / 2 ГБ
 - Автоматический чанкинг для длинных записей (>30 мин)
@@ -38,8 +38,7 @@ HF_TOKEN=hf_your_token_here
 
 > Для получения токена: https://huggingface.co/settings/tokens
 > Необходимо принять условия использования моделей:
-> - https://huggingface.co/pyannote/speaker-diarization-3.1
-> - https://huggingface.co/pyannote/segmentation-3.0
+> - https://huggingface.co/pyannote/speaker-diarization-community-1
 
 ## Использование
 
diff --git a/requirements.txt b/requirements.txt
index ef65724..481f457 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,5 +1,4 @@
 faster-whisper
 pyannote.audio
 python-dotenv
-pydub
 tqdm
diff --git a/transcriber/diarization/pyannote_engine.py b/transcriber/diarization/pyannote_engine.py
index 7ba07c1..ec95676 100644
--- a/transcriber/diarization/pyannote_engine.py
+++ b/transcriber/diarization/pyannote_engine.py
@@ -23,7 +23,7 @@ class DiarizationEngine:
     def __init__(self, hf_token: str, device: str):
         logger.info("Loading diarization pipeline on %s", device)
         self._pipeline = Pipeline.from_pretrained(
-            "pyannote/speaker-diarization-3.1",
+            "pyannote/speaker-diarization-community-1",
             token=hf_token,
         )
         self._device = torch.device(device)
@@ -57,7 +57,7 @@ class DiarizationEngine:
             diarization = self._pipeline(audio_path, hook=hook, **kwargs)
 
         turns = []
-        for turn, _, speaker in diarization.itertracks(yield_label=True):
+        for turn, speaker in diarization.exclusive_speaker_diarization:
             turns.append(SpeakerTurn(
                 start=turn.start,
                 end=turn.end,