sonusai 0.18.1__tar.gz → 0.18.2__tar.gz

{sonusai-0.18.1 → sonusai-0.18.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sonusai
-Version: 0.18.1
+Version: 0.18.2
 Summary: Framework for building deep neural network models for sound, speech, and voice AI
 Home-page: https://aaware.com
 License: GPL-3.0-only

{sonusai-0.18.1 → sonusai-0.18.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "sonusai"
-version = "0.18.1"
+version = "0.18.2"
 description = "Framework for building deep neural network models for sound, speech, and voice AI"
 authors = ["Chris Eddington <chris@aaware.com>", "Jason Calderwood <jason@aaware.com>"]
 maintainers = ["Chris Eddington <chris@aaware.com>", "Jason Calderwood <jason@aaware.com>"]

{sonusai-0.18.1 → sonusai-0.18.2}/sonusai/mixture/mixdb.py

@@ -4,7 +4,6 @@ from functools import partial
 from sqlite3 import Connection
 from sqlite3 import Cursor
 from typing import Any
-from typing import Callable
 from typing import Optional
 
 from sonusai.mixture.datatypes import AudioF
@@ -1062,13 +1061,13 @@ class MixtureDatabase:
 
     def mixids_for_speech_metadata(self,
                                    tier: str,
-                                   value: str | None,
-                                   predicate: Callable[[str], bool] = None) -> list[int]:
+                                   value: str = None,
+                                   where: str = None) -> list[int]:
         """Get a list of mixture IDs for the given speech metadata tier.
 
-        If 'predicate' is None, then include mixture IDs whose tier values are equal to the given 'value'.
-        If 'predicate' is not None, then ignore 'value' and use the given callable to determine which entries
-        to include.
+        If 'where' is None, then include mixture IDs whose tier values are equal to the given 'value'.
+        If 'where' is not None, then ignore 'value' and use the given SQL where clause to determine
+        which entries to include.
 
         Examples:
         >>> mixdb = MixtureDatabase('/mixdb_location')
@@ -1076,35 +1075,33 @@ class MixtureDatabase:
         >>> mixids = mixdb.mixids_for_speech_metadata('speaker_id', 'TIMIT_ARC0')
         Get mixutre IDs for mixtures with speakers whose speaker_ids are 'TIMIT_ARC0'.
 
-        >>> mixids = mixdb.mixids_for_speech_metadata('age', '', lambda x: int(x) < 25)
+        >>> mixids = mixdb.mixids_for_speech_metadata('age', where='age < 25')
         Get mixture IDs for mixtures with speakers whose ages are less than 25.
 
-        >>> mixids = mixdb.mixids_for_speech_metadata('dialect', '', lambda x: x in ['New York City', 'Northern'])
+        >>> mixids = mixdb.mixids_for_speech_metadata('dialect', where="dialect in ('New York City', 'Northern')")
         Get mixture IDs for mixtures with speakers whose dialects are either 'New York City' or 'Northern'.
         """
-        from .helpers import get_textgrid_tier_from_target_file
+        from sonusai import SonusAIError
 
-        if predicate is None:
-            def predicate(x: str | None) -> bool:
-                return x == value
+        if value is None and where is None:
+            raise SonusAIError('Must provide either value or where')
 
-        # First get list of matching target files
-        target_file_ids: list[int] = []
-        is_textgrid = tier in self.textgrid_metadata_tiers
-        for target_file_id, target_file in enumerate(self.target_files):
-            if is_textgrid:
-                metadata = get_textgrid_tier_from_target_file(target_file.name, tier)
-            else:
-                metadata = self.speaker(target_file.speaker_id, tier)
+        if where is None:
+            where = f"{tier} = '{value}'"
 
-            if not isinstance(metadata, list) and predicate(metadata):
-                target_file_ids.append(target_file_id + 1)
+        if tier in self.textgrid_metadata_tiers:
+            raise SonusAIError(f'TextGrid tier data, "{tier}", is not supported in mixids_for_speech_metadata().')
 
-        # Next get list of mixture IDs that contain those target files
         with self.db() as c:
-            m_ids = c.execute("SELECT mixture_id FROM mixture_target " +
-                               f"WHERE mixture_target.target_id IN ({','.join(map(str, target_file_ids))})").fetchall()
-        return [x[0] - 1 for x in m_ids]
+            speaker_ids = [speaker_id[0] for speaker_id in
+                           c.execute(f"SELECT id FROM speaker WHERE {where}").fetchall()]
+            results = c.execute(f"SELECT id FROM target_file " +
+                                f"WHERE speaker_id IN ({','.join(map(str, speaker_ids))})").fetchall()
+            target_file_ids = [target_file_id[0] for target_file_id in results]
+            results = c.execute("SELECT mixture_id FROM mixture_target " +
+                                f"WHERE mixture_target.target_id IN ({','.join(map(str, target_file_ids))})").fetchall()
+
+        return [mixture_id[0] - 1 for mixture_id in results]
 
     def mixture_all_speech_metadata(self, m_id: int) -> list[dict[str, SpeechMetadata]]:
         from .helpers import mixture_all_speech_metadata
@@ -1160,7 +1157,7 @@ class MixtureDatabase:
         )
 
         if not (metric in supported_metrics or metric.startswith('MXWER')):
-            raise ValueError(f'Unsupported metric: {metric}')
+            raise SonusAIError(f'Unsupported metric: {metric}')
 
         if not force:
             result = self.read_mixture_data(m_id, metric)