phoonnx 0.2.6a2__py3-none-any.whl → 0.2.7a1__py3-none-any.whl

phoonnx/version.py

@@ -1,8 +1,8 @@
 # START_VERSION_BLOCK
 VERSION_MAJOR = 0
 VERSION_MINOR = 2
-VERSION_BUILD = 6
-VERSION_ALPHA = 2
+VERSION_BUILD = 7
+VERSION_ALPHA = 1
 # END_VERSION_BLOCK
 
 VERSION_STR = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"

{phoonnx-0.2.6a2.dist-info → phoonnx-0.2.7a1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: phoonnx
-Version: 0.2.6a2
+Version: 0.2.7a1
 Home-page: https://github.com/TigreGotico/phoonnx
 Author: JarbasAi
 Author-email: jarbasai@mailfence.com

{phoonnx-0.2.6a2.dist-info → phoonnx-0.2.7a1.dist-info}/RECORD

@@ -2,7 +2,7 @@ phoonnx/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoonnx/config.py,sha256=DKgsU03g8jrAuMcVqbu-w3MWPXOUihFtRnavg6WGQ1Y,19983
 phoonnx/phoneme_ids.py,sha256=FiNgZwV6naEsBh6XwFLh3_FyOgPiCsK9qo7S0v-CmI4,13667
 phoonnx/util.py,sha256=XSjFEoqSFcujFTHxednacgC9GrSYyF-Il5L6Utmxmu4,25909
-phoonnx/version.py,sha256=XtQ8LLEKPJw1x3fELi5iAV6qNgCNVf_oTmwVBTXYBqA,237
+phoonnx/version.py,sha256=_KNtUP6KVTuU3xs9ekGaRTElqbX7r7YOhKIJPjmLQtc,237
 phoonnx/voice.py,sha256=Fh60o87-_sIw4zTP8NEGvEV7XzTC0FhW9Jkq7Fr60lQ,19382
 phoonnx/locale/ca/phonetic_spellings.txt,sha256=igv3t7jxLSRE5GHsdn57HOpxiWNcEmECPql6m02wbO0,47
 phoonnx/locale/en/phonetic_spellings.txt,sha256=xGQlWOABLzbttpQvopl9CU-NnwEJRqKx8iuylsdUoQA,27
@@ -63,7 +63,7 @@ phoonnx/thirdparty/tashkeel/input_id_map.json,sha256=cnpJqjx-k53AbzKyfC4GxMS771l
 phoonnx/thirdparty/tashkeel/model.onnx,sha256=UsQNQsoJT_n_B6CR0KHq_XuqXPI4jmCpzIm6zY5elV8,4788213
 phoonnx/thirdparty/tashkeel/target_id_map.json,sha256=baNAJL_UwP9U91mLt01aAEBRRNdGr-csFB_O6roh7TA,181
 phoonnx_train/export_onnx.py,sha256=UFn1id5-p9tIXxK9FvHOry59WkotNqXtupinkeKcSDs,12881
-phoonnx_train/preprocess.py,sha256=VYyU8XaCroa6tnLhQkQJ8wq1w1Gf9PPnIIM1N0PIcNE,22042
+phoonnx_train/preprocess.py,sha256=dAuHEF4-yEVhVpsVTRNat_S6_jljlwa4VpM5SZezyWc,23146
 phoonnx_train/train.py,sha256=nsINvDQ3dYvBne5UWPgLZ0a4qZFdSsOKk8HzZHGTLY4,8757
 phoonnx_train/norm_audio/__init__.py,sha256=Al_YwqMnENXRWp0c79cDZqbdd7pFYARXKxCfBaedr1c,3030
 phoonnx_train/norm_audio/trim.py,sha256=_ZsE3SYhahQSdEdBLeSwyFJGcvEbt-5E_lnWwTT4tcY,1698
@@ -83,7 +83,7 @@ phoonnx_train/vits/utils.py,sha256=exiyrtPHbnnGvcHWSbaH9-gR6srH5ZPHlKiqV2IHUrQ,4
 phoonnx_train/vits/wavfile.py,sha256=oQZiTIrdw0oLTbcVwKfGXye1WtKte6qK_52qVwiMvfc,26396
 phoonnx_train/vits/monotonic_align/__init__.py,sha256=5IdAOD1Z7UloMb6d_9NRFsXoNIjEQ3h9mvOSh_AtO3k,636
 phoonnx_train/vits/monotonic_align/setup.py,sha256=0K5iJJ2mKIklx6ncEfCQS34skm5hHPiz9vRlQEvevvY,266
-phoonnx-0.2.6a2.dist-info/METADATA,sha256=E3QnJVGElqhpgVBWDPL5ZT1PGH2fxB4lQaeXZYO99Kg,8250
-phoonnx-0.2.6a2.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
-phoonnx-0.2.6a2.dist-info/top_level.txt,sha256=ZrnHXe-4HqbOSX6fbdY-JiP7YEu2Bok9T0ji351MrmM,22
-phoonnx-0.2.6a2.dist-info/RECORD,,
+phoonnx-0.2.7a1.dist-info/METADATA,sha256=h_Ob321ENQZXjPpllqP5W3IO6ifQVgTtz8vK4AoKosE,8250
+phoonnx-0.2.7a1.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
+phoonnx-0.2.7a1.dist-info/top_level.txt,sha256=ZrnHXe-4HqbOSX6fbdY-JiP7YEu2Bok9T0ji351MrmM,22
+phoonnx-0.2.7a1.dist-info/RECORD,,

phoonnx_train/preprocess.py

@@ -377,6 +377,16 @@ def phonemize_worker(
     is_flag=True,
     help="Add diacritics to text (phonemizer specific, e.g., to denote stress).",
 )
+@click.option(
+    "--jsonl-audio-path",
+    default=None,
+    help="override audio_path base directory (everything before '/wav') in generated dataset.jsonl"
+)
+@click.option(
+    "--jsonl-audio-spec-path",
+    default=None,
+    help="override audio_norm_path/audio_spec_path base directory (everything before '/cache') in generated dataset.jsonl"
+)
 def cli(
     input_dir: Path,
     output_dir: Path,
@@ -397,6 +407,8 @@ def cli(
     skip_audio: bool,
     debug: bool,
     add_diacritics: bool,
+    jsonl_audio_path: Optional[str],
+    jsonl_audio_spec_path: Optional[str],
 ) -> None:
     """
     Preprocess a TTS dataset (e.g., LJSpeech format) for training a VITS-style model.
@@ -609,6 +621,17 @@ def cli(
                 _LOGGER.warning("Skipping utterance with invalid phoneme_ids before writing: %s", utt.audio_path)
                 continue
 
+            # apply path overrides if needed
+            # this allows pre-processing the dataset in one system and then train in other
+            if jsonl_audio_path:
+                base_path, fname = str(utt.audio_path).split("/wav/")
+                utt.audio_path = Path(f"{jsonl_audio_path}/wav/{fname}")
+            if jsonl_audio_spec_path:
+                base_path, fname = str(utt.audio_norm_path).split("/cache/")
+                utt.audio_norm_path = Path(f"{jsonl_audio_spec_path}/cache/{fname}")
+                base_path, fname = str(utt.audio_spec_path).split("/cache/")
+                utt.audio_spec_path = Path(f"{jsonl_audio_spec_path}/cache/{fname}")
+
             json.dump(
                 utt.asdict(),
                 dataset_file,