add frm expand and trim in kaldi_io

2 files changed, 47 insertions, 18 deletions

diff --git a/htk_io/init.lua b/htk_io/init.lua
index abaa794..b360b67 100644
--- a/htk_io/init.lua
+++ b/htk_io/init.lua
@@ -14,12 +14,14 @@ function TNetReader:__init(global_conf, reader_conf)
                                           reader_conf.conf_file,
                                           reader_conf.frm_ext)
     self.lab_repo = {}
-    for id, mlf_spec in pairs(reader_conf.mlfs) do
-        self.lab_repo[id] = nerv.TNetLabelRepo(mlf_spec.file,
-                                            mlf_spec.format,
-                                            mlf_spec.format_arg,
-                                            mlf_spec.dir,
-                                            mlf_spec.ext)
+    if reader_conf.mlfs then
+        for id, mlf_spec in pairs(reader_conf.mlfs) do
+            self.lab_repo[id] = nerv.TNetLabelRepo(mlf_spec.file,
+                                                mlf_spec.format,
+                                                mlf_spec.format_arg,
+                                                mlf_spec.dir,
+                                                mlf_spec.ext)
+        end
     end
 end
 
diff --git a/kaldi_io/init.lua b/kaldi_io/init.lua
index 930705e..3fc5b10 100644
--- a/kaldi_io/init.lua
+++ b/kaldi_io/init.lua
@@ -13,20 +13,24 @@ function KaldiReader:__init(global_conf, reader_conf)
     self.feat_repo = nerv.KaldiFeatureRepo(reader_conf.feature_rspecifier)
 
     self.lab_repo = {}
-    for id, mlf_spec in pairs(reader_conf.mlfs) do
-        if mlf_spec.format == nil then
-            nerv.error("format spec is expected for label %s", id)
+    if reader_conf.mlfs then
+        for id, mlf_spec in pairs(reader_conf.mlfs) do
+            if mlf_spec.format == nil then
+                nerv.error("format spec is expected for label %s", id)
+            end
+            self.lab_repo[id] = nerv.KaldiLabelRepo(mlf_spec.targets_rspecifier,
+                                                mlf_spec.format)
         end
-        self.lab_repo[id] = nerv.KaldiLabelRepo(mlf_spec.targets_rspecifier,
-                                            mlf_spec.format)
     end
     self.lookup_repo = {}
-    for id, lookup_spec in pairs(reader_conf.lookup) do
-        if lookup_spec.map_rspecifier == nil then
-            nerv.error("map spec is expected for lookup %s", id)
+    if reader_conf.lookup then
+        for id, lookup_spec in pairs(reader_conf.lookup) do
+            if lookup_spec.map_rspecifier == nil then
+                nerv.error("map spec is expected for lookup %s", id)
+            end
+            self.lookup_repo[id] = nerv.KaldiLookupFeatureRepo(lookup_spec.targets_rspecifier,
+                                                               lookup_spec.map_rspecifier)
         end
-        self.lookup_repo[id] = nerv.KaldiLookupFeatureRepo(lookup_spec.targets_rspecifier,
-                                                           lookup_spec.map_rspecifier)
     end
 end
 
@@ -36,8 +40,31 @@ function KaldiReader:get_data()
     end
     local res = {}
     -- read Kaldi feature
-    local feat_utter = self.feat_repo:cur_utter(self.debug)
-    -- global transf
+    local raw = self.gconf.cumat_type.new_from_host(self.feat_repo:cur_utter(self.debug))
+    local rearranged
+    if self.frm_ext and self.frm_ext > 0 then
+        local step = self.frm_ext * 2 + 1
+        -- expand the feature
+        local expanded = self.gconf.cumat_type(raw:nrow(), raw:ncol() * step)
+        expanded:expand_frm(raw, self.frm_ext)
+        -- rearrange the feature (``transpose'' operation in TNet)
+        if self.gconf.rearrange then
+            rearranged = expanded:create()
+            rearranged:rearrange_frm(expanded, step)
+        else
+            rearranged = expanded
+        end
+    else
+        rearranged = raw
+    end
+    local feat_utter
+    if self.gconf.frm_trim then
+        feat_utter = self.gconf.mmat_type(rearranged:nrow() - self.gconf.frm_trim * 2, rearranged:ncol())
+        rearranged:copy_toh(feat_utter, self.gconf.frm_trim, rearranged:nrow() - self.gconf.frm_trim)
+    else
+        feat_utter = self.gconf.mmat_type(rearranged:nrow(), rearranged:ncol())
+        rearranged:copy_toh(feat_utter)
+    end
     res[self.feat_id] = feat_utter
     -- add corresponding labels
     for id, repo in pairs(self.lab_repo) do