refactor: split tts.py into per-backend packages

1 files changed, 14 insertions, 7 deletions

diff --git a/app/converter/converter.py b/app/converter/converter.py
index 1abc85c..48d4987 100644
--- a/app/converter/converter.py
+++ b/app/converter/converter.py
@@ -15,7 +15,7 @@ from typing import Callable, Dict, List, Optional, Tuple
 
 from . import audio, chunking, config, cover, extractors
 from .audio import TrackMeta
-from .tts import (
+from .clients import (
     BACKENDS,
     BACKEND_AUDIOCPP,
     BACKEND_FASTER,
@@ -228,7 +228,8 @@ class AudiobookConverter:
         if backend == BACKEND_FASTER:
             # The faster backend always voice-clones using a reference voice
             # configured on the server, so no local reference audio is needed.
-            self.tts = FasterTTSClient(voice=voice, api_url=api_url,
+            self.tts = FasterTTSClient(chunks_dir=CHUNKS_FOLDER,
+                                       voice=voice, api_url=api_url,
                                        quiet=quiet)
         elif backend == BACKEND_AUDIOCPP:
             # --voice picks the voice: a built-in speaker name on the
@@ -237,13 +238,15 @@ class AudiobookConverter:
             # multi-model servers; instructions describe or style the
             # voice, request_options pass per-model controls through to
             # the server.
-            self.tts = AudioCppTTSClient(voice=voice, language=self.language,
+            self.tts = AudioCppTTSClient(chunks_dir=CHUNKS_FOLDER,
+                                         voice=voice, language=self.language,
                                          model_id=model_id,
                                          instructions=instructions,
                                          request_options=self.request_options,
                                          api_url=api_url, quiet=quiet)
         else:
             self.tts = QwenTTSClient(
+                chunks_dir=CHUNKS_FOLDER,
                 voice_mode=voice_mode,
                 voice_clone_ref_audio=voice_clone_ref_audio,
                 voice_clone_ref_text=voice_clone_ref_text,
@@ -398,7 +401,7 @@ class AudiobookConverter:
             self.current_outputs = []
             # Start from a clean scratch folder so a previous crash can never
             # affect this run
-            audio.cleanup_chunks()
+            audio.cleanup_chunks(CHUNKS_FOLDER)
 
             logger.info("Extracting text...")
             book = extractors.extract_book(file_path)
@@ -476,7 +479,7 @@ class AudiobookConverter:
             return False
         finally:
             # Always cleanup, even on failure or interrupt
-            audio.cleanup_chunks()
+            audio.cleanup_chunks(CHUNKS_FOLDER)
 
     def _convert_m4b_with_chapters(self, sections, stem: str, start_time: float,
                                    meta: Optional[TrackMeta] = None,
@@ -519,7 +522,9 @@ class AudiobookConverter:
             return False
 
         output_path = AUDIOBOOKS_FOLDER / f"{stem}.{self.output_format}"
-        if not audio.combine_chapters_to_m4b(chapter_files, titles, output_path, speed=self.speed,
+        if not audio.combine_chapters_to_m4b(chapter_files, titles, output_path,
+                                             chunks_dir=CHUNKS_FOLDER,
+                                             speed=self.speed,
                                              meta=meta, cover=cover):
             return False
         duration = time.time() - start_time
@@ -653,7 +658,9 @@ class AudiobookConverter:
                              successful_chunks, total_chunks)
                 return False
 
-            success = audio.combine_chunks(total_chunks, output_path, chunk_results=results,
+            success = audio.combine_chunks(total_chunks, output_path,
+                                           chunk_results=results,
+                                           chunks_dir=CHUNKS_FOLDER,
                                            speed=speed, output_format=output_format,
                                            intermediate=chapter is not None,
                                            meta=meta, cover=cover)