1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
|
"""Reference-.wav transcription planning and execution."""
import argparse
from pathlib import Path
from typing import Dict, List, Optional, Tuple
from backends.common import (PROMPT_TEXT_FILENAME, find_wav_files,
read_prompt_text)
from converter.clients import (transcribe_reference_audio,
whisper_backend_available)
def transcribe_wav_dir(wav_files: list, whisper_model: str,
cancel=None) -> Dict[str, str]:
"""Transcribe each wav file and return a mapping of stem -> transcript.
CANCEL (a ``threading.Event``) is checked between files so the in-TUI
task view can stop a long transcription early.
"""
transcripts: Dict[str, str] = {}
for wav_file in wav_files:
if cancel is not None and cancel.is_set():
print("[INFO] Transcription cancelled")
break
name = wav_file.stem
print(f"[INFO] Transcribing {wav_file.name} (voice '{name}')...")
text = transcribe_reference_audio(str(wav_file), model_name=whisper_model)
if text:
print(f"[OK] {name}: {text}")
else:
print(f"[WARNING] No transcript for '{name}'; cloning works best "
"with an accurate transcript — consider editing prompt_text "
"by hand before starting the server")
transcripts[name] = text or ""
return transcripts
def print_empty_transcript_warning(transcripts: Dict[str, str]) -> None:
"""Print a loud, final warning for voices whose transcript is empty."""
empty = sorted(name for name, text in transcripts.items() if not text)
if not empty:
return
bar = "=" * 70
print()
print(bar)
print("[WARNING] MANUAL TRANSCRIPTION REQUIRED")
print(bar)
listing = " - " + "\n - ".join(empty) if len(empty) > 1 else f" - {empty[0]}"
print(f"The following voice(s) have an EMPTY transcript in prompt_text:\n"
f"{listing}")
print("Those voices will NOT work until you add an accurate transcript.")
print(f"Edit {PROMPT_TEXT_FILENAME} in your voice directory and fill in the "
"text after '|' for each voice above.")
print(bar)
def _transcribe(args: argparse.Namespace, plan: Optional[dict],
cancel=None) -> Tuple[Dict[str, str], bool]:
"""Transcribe the wav directory into a stem -> transcript mapping.
Returns the mapping and a flag indicating whether it should be written to
prompt_text (False when an existing, complete prompt_text is kept as-is).
PLAN is always pre-collected — by the TUI setup form (mode "all",
"missing" or "keep") or by _flag_plan for a non-interactive run — so no
questions are asked here; a None PLAN defaults to "transcribe everything".
CANCEL is checked between files.
"""
wav_files = find_wav_files(args.input_dir)
if not wav_files:
print(f"[WARNING] No .wav files found in {args.input_dir}; writing the "
"config without a voice_dir")
return {}, False
prompt_path = args.input_dir / PROMPT_TEXT_FILENAME
existing = dict((plan or {}).get("existing") or {})
mode = plan["mode"] if plan else "all"
if mode == "keep":
print(f"[INFO] Kept existing {prompt_path}; all voices were "
"already transcribed, nothing new to transcribe")
return existing, False
if whisper_backend_available() is None:
print("[WARNING] Neither faster_whisper nor whisper was found, so "
"reference .wav files cannot be transcribed automatically and "
"every transcript will be empty.")
print(" Install whisper (or faster_whisper) in your "
"audiobook environment to transcribe automatically; otherwise "
"transcripts must be added by hand (see the warning at the end).")
if plan["mode"] == "missing":
new_transcripts = transcribe_wav_dir(plan["missing"], args.whisper_model,
cancel=cancel)
transcripts = dict(existing)
transcripts.update(new_transcripts)
else:
transcripts = transcribe_wav_dir(wav_files, args.whisper_model,
cancel=cancel)
return transcripts, True
def _flag_plan(wav_files: list, prompt_path: Path, force: bool) -> dict:
"""Build a transcription plan for a non-interactive (flag-only) run.
With --force everything is re-transcribed; otherwise an existing
prompt_text is reused and only voices with an empty transcript are
re-transcribed, mirroring what the TUI confirms interactively.
"""
if prompt_path.exists() and not force:
existing = read_prompt_text(prompt_path)
missing = [wav for wav in wav_files
if not existing.get(wav.stem, "").strip()]
if not missing:
return {"mode": "keep", "missing": [], "existing": existing}
return {"mode": "missing", "missing": missing, "existing": existing}
return {"mode": "all", "missing": [], "existing": {}}
|
