Mid indexing problem fix

src/sppysound/analysis/Analysis.py

@@ -75,7 +75,7 @@ class Analysis(object):
                 # be saved in the HDF5 file
                 data_dict, attrs_dict = self.hdf5_dataset_formatter(*args, **kwargs)
                 for key, value in data_dict.iteritems():
-                    self.analysis.create_dataset(key, data=value)
+                    self.analysis.create_dataset(key, data=value, dtype=np.float, chunks=True)
                 for key, value in attrs_dict.iteritems():
                     self.analysis.attrs[key] = value
 
@@ -145,6 +145,9 @@ class Analysis(object):
         }
         output = np.empty(len(selection))
 
+        if not selection.size:
+            # TODO: Add warning here
+            return np.nan
         # For debugging apply_along_axis:
         #for ind, i in enumerate(selection):
         #    output[ind] = self.formatter_func(i, frames, valid_inds, formatter=format_style_dict[format])

src/sppysound/analysis/F0Analysis.py

@@ -251,17 +251,24 @@ class F0Analysis(Analysis):
         }
 
         # For debugging apply along axis:
+        if not selection.size:
+            # TODO: Add warning here
+            return np.nan
+
         #for ind, i in enumerate(selection):
         #    output[ind] = self.formatter_func(i, frames, valid_inds, harm_ratio, formatter=format_style_dict[format])
 
-        output = np.apply_along_axis(
-            self.formatter_func,
-            1,
-            selection,
-            frames,
-            valid_inds,
-            formatter=format_style_dict[format]
-        )/self.nyquist_rate
+        try:
+            output = np.apply_along_axis(
+                self.formatter_func,
+                1,
+                selection,
+                frames,
+                valid_inds,
+                formatter=format_style_dict[format]
+            )/self.nyquist_rate
+        except IndexError:
+            pdb.set_trace()
 
         return output
 

src/sppysound/analysis/F0HarmRatioAnalysis.py

@@ -82,6 +82,10 @@ class F0HarmRatioAnalysis(Analysis):
         #for ind, i in enumerate(selection):
         #    output[ind] = self.formatter_func(i, frames, valid_inds, harm_ratio, formatter=format_style_dict[format])
 
+        if not selection.size:
+            # TODO: Add warning here
+            return np.nan
+
         output = np.apply_along_axis(
             self.formatter_func,
             1,

src/sppysound/concatenator.py

@@ -264,7 +264,6 @@ def main():
         log_filename=modpath,
         logger_filelevel=args.verbose
     )
-    pdb.set_trace()
 
     # Create/load a pre-existing source database
     source_db = AudioDatabase(

src/sppysound/database.py

@@ -410,24 +410,33 @@ class Matcher:
                 target_data, s = target_entry.analysis_data_grains(target_times, analysis, format=analysis_formatting)
                 all_target_analyses[i] = target_data
 
+
             imp = Imputer(axis=1)
+            nan_columns = np.all(np.isnan(all_target_analyses), axis=1)
+            all_target_analyses[nan_columns, :] = 0.
             # Impute values for Nans
             all_target_analyses = imp.fit_transform(all_target_analyses)
 
             for sind, source_entry in enumerate(self.source_db.analysed_audio):
+                self.logger.info("K-d Tree Matching: {0} to {1}".format(source_entry.name, target_entry.name))
                 # Create an array of grain times for source sample
                 source_times = source_entry.generate_grain_times(grain_size, overlap, save_times=True)
+                if not source_times.size:
+                    continue
+
                 all_source_analyses = np.empty((len(self.matcher_analyses), source_times.shape[0]))
 
+
                 for i, analysis in enumerate(self.matcher_analyses):
                     analysis_formatting = self.analysis_dict[analysis]
 
                     source_data, s = source_entry.analysis_data_grains(source_times, analysis, format=analysis_formatting)
                     all_source_analyses[i] = source_data
 
-                self.logger.info("K-d Tree Matching: {0} to {1}".format(source_entry.name, target_entry.name))
 
                 # Impute values for Nans
+                nan_columns = np.all(np.isnan(all_source_analyses), axis=1)
+                all_source_analyses[nan_columns, :] = 0.
                 all_source_analyses = imp.fit_transform(all_source_analyses)
 
                 source_tree = spatial.cKDTree(all_source_analyses.T, leafsize=100)
@@ -448,6 +457,7 @@ class Matcher:
                 match_vals = best_match_vals[:, :self.match_quantity]
 
             match_grain_inds = self.calculate_db_inds(match_indexes, source_sample_indexes)
+            pdb.set_trace()
 
             datafile_path = ''.join(("match/", target_entry.name))
             try:
@@ -787,6 +797,7 @@ class Synthesizer:
                     # from available matches.
                     match_index = np.random.randint(matches.shape[0])
                     match_db_ind, match_grain_ind = matches[match_index]
+                    pdb.set_trace()
                     with self.match_db.analysed_audio[match_db_ind] as match_sample:
                         match_sample.generate_grain_times(match_grain_size, match_overlap, save_times=True)