nkululeko 0.95.8__py3-none-any.whl → 0.96.0__py3-none-any.whl

nkululeko/autopredict/ap_translate.py

@@ -0,0 +1,39 @@
+"""A translator for text.
+
+Currently based on google translate.
+"""
+
+from nkululeko.utils.util import Util
+
+
+class TextTranslator:
+    """Translator.
+
+    translate text with the google translate model
+    """
+
+    def __init__(self, df, util=None):
+        self.df = df
+        if util is not None:
+            self.util = util
+        else:
+            # create a new util instance
+            # this is needed to access the config and other utilities
+            # in the autopredict module
+            self.util = Util("translator")
+            
+        self.language = self.util.config_val("PREDICT", "target_language", "en")
+        from nkululeko.autopredict.google_translator import GoogleTranslator
+        self.translator = GoogleTranslator(
+            language=self.language,
+            util=self.util,
+        )
+
+    def predict(self, split_selection):
+        self.util.debug(f"translating text for {split_selection} samples")
+        df = self.translator.translate_index(
+            self.df
+        )
+        return_df = self.df.copy()
+        return_df[self.language] = df[self.language].values
+        return return_df

nkululeko/autopredict/google_translator.py

@@ -0,0 +1,63 @@
+import os
+
+import pandas as pd
+import torch
+from tqdm import tqdm
+
+import asyncio
+from googletrans import Translator
+
+import audeer
+import audiofile
+
+from nkululeko.utils.util import Util
+
+import httpx
+
+class GoogleTranslator:
+    def __init__(self, language="en", util=None):
+        self.language = language
+        self.util = util
+
+    async def translate_text(self, text):
+        async with Translator() as translator:
+            result = translator.translate(text, dest="en")
+            return (await result).text
+
+    def translate_index(self, df:pd.DataFrame) ->  pd.DataFrame:
+        """Transcribe the audio files in the given index.
+
+        :param index: Index containing tuples of (file, start, end).
+        :return: DataFrame with transcriptions indexed by the original index.
+        :rtype: pd.DataFrame
+        """
+        file_name = ""
+        seg_index = 0
+        translations = []
+        translator_cache = audeer.mkdir(
+            audeer.path(self.util.get_path("cache"), "translations"))
+        file_name = ""
+        for idx, row in tqdm(df.iterrows(), total=len(df)):
+            file = idx[0]
+            start = idx[1]
+            end = idx[2]
+            if file != file_name:
+                file_name = file
+                seg_index = 0
+            cache_name = audeer.basename_wo_ext(file)+str(seg_index)
+            cache_path = audeer.path(translator_cache, cache_name + ".json")
+            if os.path.isfile(cache_path):
+                translation = self.util.read_json(cache_path)["translation"]
+            else:
+                text = row['text']
+                translation = asyncio.run(self.translate_text(text))
+                self.util.save_json(cache_path, 
+                                {"translation": translation, 
+                                 "file": file, 
+                                 "start": start.total_seconds(), 
+                                 "end": end.total_seconds()})
+            translations.append(translation)
+            seg_index += 1
+
+        df = pd.DataFrame({self.language:translations}, index=df.index)
+        return df

nkululeko/constants.py

@@ -1,2 +1,2 @@
-VERSION="0.95.8"
+VERSION="0.96.0"
 SAMPLING_RATE = 16000

nkululeko/experiment.py

@@ -574,6 +574,11 @@ class Experiment:
 
                 predictor = TextPredictor(df, self.util)
                 df = predictor.predict(sample_selection)
+            elif target == "translation":
+                from nkululeko.autopredict.ap_translate import TextTranslator
+
+                predictor = TextTranslator(df, self.util)
+                df = predictor.predict(sample_selection)
             elif target == "arousal":
                 from nkululeko.autopredict.ap_arousal import ArousalPredictor
 

nkululeko/feat_extract/feats_bert.py

@@ -0,0 +1,105 @@
+import os
+
+import pandas as pd
+from tqdm import tqdm
+import transformers
+import torch 
+from transformers import BertTokenizer, BertModel
+
+from nkululeko.feat_extract.featureset import Featureset
+import nkululeko.glob_conf as glob_conf
+
+
+class Bert(Featureset):
+    """Class to extract bert embeddings"""
+
+    def __init__(self, name, data_df, feat_type):
+        """Constructor.
+
+        If_train is needed to distinguish from test/dev sets,
+        because they use the codebook from the training
+        """
+        super().__init__(name, data_df, feat_type)
+        cuda = "cuda" if torch.cuda.is_available() else "cpu"
+        self.device = self.util.config_val("MODEL", "device", cuda)
+        self.model_initialized = False
+        if feat_type == "bert":
+            self.feat_type = "bert-base-uncased"
+        else:
+            self.feat_type = feat_type
+
+    def init_model(self):
+        # load model
+        self.util.debug(f"loading {self.feat_type} model...")
+        model_path = self.util.config_val(
+            "FEATS", "bert.model", f"google-bert/{self.feat_type}"
+        )
+        config = transformers.AutoConfig.from_pretrained(model_path)
+        layer_num = config.num_hidden_layers
+        hidden_layer = int(self.util.config_val("FEATS", "bert.layer", "0"))
+        config.num_hidden_layers = layer_num - hidden_layer
+        self.util.debug(f"using hidden layer #{config.num_hidden_layers}")
+
+        self.tokenizer = BertTokenizer.from_pretrained(model_path)
+        self.model = BertModel.from_pretrained(model_path, config=config).to(
+            self.device
+        )
+        print(f"initialized {self.feat_type} model on {self.device}")
+        self.model.eval()
+        self.model_initialized = True
+
+    def extract(self):
+        """Extract the features or load them from disk if present."""
+        store = self.util.get_path("store")
+        storage = os.path.join(store, f"{self.name}.pkl")
+        extract = self.util.config_val("FEATS", "needs_feature_extraction", False)
+        no_reuse = eval(self.util.config_val("FEATS", "no_reuse", "False"))
+        if extract or no_reuse or not os.path.isfile(storage):
+            if not self.model_initialized:
+                self.init_model()
+            self.util.debug(
+                f"extracting {self.feat_type} embeddings, this might take a while..."
+            )
+            emb_series = pd.Series(index=self.data_df.index, dtype=object)
+            for idx, row in tqdm(self.data_df.iterrows(), total=len(self.data_df)):
+                file = idx[0]
+                text = row['text']
+                emb = self.get_embeddings(text, file)
+                emb_series[idx] = emb
+            # print(f"emb_series shape: {emb_series.shape}")
+            self.df = pd.DataFrame(emb_series.values.tolist(), index=self.data_df.index)
+            # print(f"df shape: {self.df.shape}")
+            self.df.to_pickle(storage)
+            try:
+                glob_conf.config["DATA"]["needs_feature_extraction"] = "false"
+            except KeyError:
+                pass
+        else:
+            self.util.debug(f"reusing extracted {self.feat_type} embeddings")
+            self.df = pd.read_pickle(storage)
+            if self.df.isnull().values.any():
+                self.util.error(
+                    f"got nan: {self.df.shape} {self.df.isnull().sum().sum()}"
+                )
+
+    def get_embeddings(self, text, file):
+        r"""Extract embeddings from raw audio signal."""
+        try:
+            with torch.no_grad():
+                inputs = self.tokenizer(text, return_tensors="pt")
+                outputs = self.model(**inputs)
+                # mean pooling 
+                y = torch.mean(outputs[0], dim=1)
+                y = y.ravel()
+        except RuntimeError as re:
+            print(str(re))
+            self.util.error(f"couldn't extract file: {file}")
+            y = None
+        if y is None:
+            return None
+        return y.detach().cpu().numpy()
+
+    def extract_sample(self, text):
+        self.init_model()
+        feats = self.get_embeddings(text, "no file")
+        return feats

nkululeko/feature_extractor.py

@@ -80,7 +80,7 @@ class FeatureExtractor:
             return MLD_set
 
         elif feats_type.startswith(
-            ("wav2vec2", "hubert", "wavlm", "spkrec", "whisper", "ast", "emotion2vec")
+            ("bert", "wav2vec2", "hubert", "wavlm", "spkrec", "whisper", "ast", "emotion2vec")
         ):
             return self._get_feat_extractor_by_prefix(feats_type)
 

{nkululeko-0.95.8.dist-info → nkululeko-0.96.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nkululeko
-Version: 0.95.8
+Version: 0.96.0
 Summary: Machine learning audio prediction experiments based on templates
 Home-page: https://github.com/felixbur/nkululeko
 Author: Felix Burkhardt

{nkululeko-0.95.8.dist-info → nkululeko-0.96.0.dist-info}/RECORD

@@ -4,16 +4,16 @@ nkululeko/aug_train.py,sha256=wpiHCJ7zsW38kumg3ypwXZe2HQrhUblAnv7P2QeJnAc,3525
 nkululeko/augment.py,sha256=3RzaxB3gRxovgJVjHXi0glprW01J7RaHhUkqotW2T3U,2955
 nkululeko/balance.py,sha256=r7opXbrqAipm2euPPaOmLlA5J10p2bHQgO5kWk2x9ro,8702
 nkululeko/cacheddataset.py,sha256=XFpWZmbJRg0pvhnIgYf0TkclxllD-Fctu-Ol0PF_00c,969
-nkululeko/constants.py,sha256=9cVRluLqakiiCBPe3kAeJizsgpn2LUbgdAs0Y9scIEM,39
+nkululeko/constants.py,sha256=dv32YQyIGpLW6TD10egbPsxz2wdIYaw29xUA2K28xt4,39
 nkululeko/demo-ft.py,sha256=iD9Pzp9QjyAv31q1cDZ75vPez7Ve8A4Cfukv5yfZdrQ,770
 nkululeko/demo.py,sha256=tu7Al2l5MCLVegkDC-NE2wcuc_YE7NRbgOlPW3yhGEs,4940
 nkululeko/demo_feats.py,sha256=BvZjeNFTlERIRlq34OHM4Z96jdDQAhB01BGQAUcX9dM,2026
 nkululeko/demo_predictor.py,sha256=lDF-xOxRdEAclOmbepAYg-BQXQdGkHfq2n74PTIoop8,4872
 nkululeko/ensemble.py,sha256=71V-rre61H3J4sh7lu-OTo4I2_g7mm_rQxwW1ARDHgY,12782
-nkululeko/experiment.py,sha256=BAc220lktt_tvifl-m-ZIPO7Nwi-HzDBNyTfjPDbQkE,38397
+nkululeko/experiment.py,sha256=TG9G9kSETT_R8d92aRKMMsb0HRGyM_GBFHBsU9A6ppw,38633
 nkululeko/explore.py,sha256=PjNcLuPdvWqCqYXUvGhd0hBijIhzdyi3ED1RF6o5Gjk,4212
 nkululeko/export.py,sha256=U-V4acxtuL6qKt6oAsVcM5TTeWogYUJ3GU-lA6rq6d4,4336
-nkululeko/feature_extractor.py,sha256=CsKmBoxwNClRGu20ox_eCxMG4u_1OH8Y83FYw7GfUwA,4230
+nkululeko/feature_extractor.py,sha256=d3G42OOh315Aho-yLaFT739U0UI8otiB1I4ZksK8kfg,4238
 nkululeko/file_checker.py,sha256=xJY0Q6w47pnmgJVK5rcAKPYBrCpV7eBT4_3YBzTx-H8,3454
 nkululeko/filter_data.py,sha256=4sGrKvMZ_hLnJPrHm_CqjDPKIRV8REWoT7nfSYGXbwo,7305
 nkululeko/fixedsegment.py,sha256=Tb92QiuiyMsOO3WRWwuGjZGibS8hbHHCrcWAXGk7g04,2868
@@ -52,8 +52,10 @@ nkululeko/autopredict/ap_sid.py,sha256=b_JwVWlqcwdC7acU9Q7mExuOJKUn6qdlmQTm8pmmp
 nkululeko/autopredict/ap_snr.py,sha256=cjc0pUsCN_RacTw1UBR2cGY9t_um48f2cjo3QJDn7bw,1111
 nkululeko/autopredict/ap_stoi.py,sha256=csv9qCcRmieHAhypszqGoGt9r3biM8IYPgcTwp9GIFM,1188
 nkululeko/autopredict/ap_text.py,sha256=zaz9qIg90-ghZhBe1ka0HoUnap6s6RyopUKoCpttHOU,1333
+nkululeko/autopredict/ap_translate.py,sha256=3yxNQmysAB3GP84YjFg_9Wc5yz0iXZXxrL5VZpEyyiI,1138
 nkululeko/autopredict/ap_valence.py,sha256=9S06SpO_zXKSpkf0InHYYXZcD9HDGoCJ6UPkn__eBAg,1027
 nkululeko/autopredict/estimate_snr.py,sha256=1k9-XadABudnsNOeFZD_Fg0E64-GUQVS7JEp82MLQS4,4995
+nkululeko/autopredict/google_translator.py,sha256=at5v7vWbP9UwXDVqVCqai4SKz-XuazpeeFsptSGrBUU,2114
 nkululeko/autopredict/whisper_transcriber.py,sha256=DWDvpRaV5KmUF18ojPEvxnVXm_h_nWyY-TfW2Ngd5N8,2941
 nkululeko/autopredict/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nkululeko/autopredict/tests/test_whisper_transcriber.py,sha256=ilas6j3OUvq_xnQCRZgytQCtyrpNU6tvG5a8kPvVKBQ,5085
@@ -67,6 +69,7 @@ nkululeko/feat_extract/feats_analyser.py,sha256=lodim7qQ8M7c3iMeJ5bHQ-nCy9Cehx1X
 nkululeko/feat_extract/feats_ast.py,sha256=w62xEoLiFtU-rj6SXkqXAktmoFaXcAcAWpUyEjp8JWo,4652
 nkululeko/feat_extract/feats_auddim.py,sha256=CGLp_aYhudfwoU5522vjrvjPxfZcyw593A8xLjYefV8,3134
 nkululeko/feat_extract/feats_audmodel.py,sha256=OsZyB1rdcG0Fai2gAwBlbuubmWor1_-P4IDkZLqgPKE,3161
+nkululeko/feat_extract/feats_bert.py,sha256=KgWLYLA11e86ubY1KtUk74QGOrZaiUongn-2LKWyf1M,4114
 nkululeko/feat_extract/feats_clap.py,sha256=1tttpfm2SJmQgYm2u8eUVpDiDOpWdKqFChpY3ZZokNs,3395
 nkululeko/feat_extract/feats_emotion2vec.py,sha256=LnV8xEg7L7HIDqz0ulqUNoaAHBU0d5gyQPb2_32T_18,9694
 nkululeko/feat_extract/feats_hubert.py,sha256=F3vrPCkx8EimJjFWYCZ7Yg9uo1G3NjYt4UKrGIUev8k,5172
@@ -134,9 +137,9 @@ nkululeko/utils/files.py,sha256=SrrYaU7AB80MZHiV1jcB0h_zigvYLYgSVNTXV4ao38g,4593
 nkululeko/utils/stats.py,sha256=3Fyx8q8BSKYmiufT6OkRug9RATWmGrr9BaX_y8jziWo,3074
 nkululeko/utils/unzip.py,sha256=G68f5120TjwACZC3bQcneMniddnwubPbBdMc2L5KBOo,1206
 nkululeko/utils/util.py,sha256=s7Hd7Ju1r3_WCw8gLD9YK4O6k3S_WhFcN2-XZBSctSM,18705
-nkululeko-0.95.8.dist-info/licenses/LICENSE,sha256=0zGP5B_W35yAcGfHPS18Q2B8UhvLRY3dQq1MhpsJU_U,1076
-nkululeko-0.95.8.dist-info/METADATA,sha256=ye2EpYXgbMMRD1L1KnUqjbFO5W7--glJKj-1yBmHCX8,21998
-nkululeko-0.95.8.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nkululeko-0.95.8.dist-info/entry_points.txt,sha256=lNTkFEdh6Kjo5o95ZAWf_0Lq-4ztGoAoMVSDuPtuyS0,442
-nkululeko-0.95.8.dist-info/top_level.txt,sha256=bf1k1YKkqcXemNX_cUgoyKqQ3_GVErPqAY-53J36jkM,19
-nkululeko-0.95.8.dist-info/RECORD,,
+nkululeko-0.96.0.dist-info/licenses/LICENSE,sha256=0zGP5B_W35yAcGfHPS18Q2B8UhvLRY3dQq1MhpsJU_U,1076
+nkululeko-0.96.0.dist-info/METADATA,sha256=uwhWqKjvfyvV6UHK4v_2H7WusIw6Bei_7i03RWVHWHE,21998
+nkululeko-0.96.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nkululeko-0.96.0.dist-info/entry_points.txt,sha256=lNTkFEdh6Kjo5o95ZAWf_0Lq-4ztGoAoMVSDuPtuyS0,442
+nkululeko-0.96.0.dist-info/top_level.txt,sha256=bf1k1YKkqcXemNX_cUgoyKqQ3_GVErPqAY-53J36jkM,19
+nkululeko-0.96.0.dist-info/RECORD,,