clean up gemfie

Author: taf2

Gemfile

@@ -2,4 +2,4 @@ source :rubygems
 
 gem 'curb'
 gem 'json'
-gem 'speech2text', :path => ENV["HOME"] + "/google-speech-to-text/"
+gem 'speech2text', '0.3.4' #:path => ENV["HOME"] + "/google-speech-to-text/"

Gemfile.lock

@@ -1,20 +1,16 @@
-PATH
-  remote: /Users/taf2/google-speech-to-text
-  specs:
-    speech2text (0.3.4)
-      curb
-      json
-
 GEM
   remote: http://rubygems.org/
   specs:
     curb (0.7.15)
     json (1.5.1)
+    speech2text (0.3.4)
+      curb
+      json
 
 PLATFORMS
   ruby
 
 DEPENDENCIES
   curb
   json
-  speech2text!
+  speech2text (= 0.3.4)

bin/wav2txt

@@ -15,7 +15,7 @@ end
 
 input = ARGV[0]
 
-system("./splitter -i #{input} -t 15 -d 0.3 -D 10")
+system("./splitter -i #{input} -t 15 -d 0.2 -D 9")
 path = File.dirname(input)
 
 results = []
@@ -35,13 +35,51 @@ chunks = Dir["#{path}/#{root_name}*.wav.chunk*"].sort do|a, b|
 end
 
 # now attempt to combine chunks to be as close to 10 seconds as possible
-chunks.each do|chunk|
+accom = Speech::AudioInspector::Duration.new("00:00:00.00")
+merge_set = [] # array of [chunk,duration]
+merged_chunks = []
+chunks.each_with_index do|chunk, i|
   inspector = Speech::AudioInspector.new(chunk)
-  puts inspector.duration.inspect
+  stamp = inspector.duration
+  accom = accom + stamp
+
+  puts "check: #{accom}, #{accom.to_f}"
+  merge_set << [chunk, stamp]
+
+  if accom.to_f > 9.0
+    last = merge_set.pop
+
+    accom = Speech::AudioInspector::Duration.new("00:00:00.00")
+    merge_set.each {|s| accom = accom + s.last }
+    merged_list = merge_set.map {|c| c.first }.join(' ')
+    merge_name = "merged_chunk#{i}.wav"
+    puts "\tmerge(#{accom.to_s}): #{merge_set.inspect} as #{merge_name}"
+    if merge_set.size == 1
+      system("cp #{merged_list} #{merge_name}")
+    else
+      system("./merge-wave #{merged_list} #{merge_name}")
+    end
+    merged_chunks << merge_name
+
+    merge_set = [last]
+    accom = Speech::AudioInspector::Duration.from_seconds(last.last.to_f)
+  end
+end
+
+if !merge_set.empty?
+  merged_list = merge_set.map {|c| c.first }.join(' ')
+  merge_name = "merged_chunk_last.wav"
+  system("./merge-wave #{merged_list} #{merge_name}")
+  merged_chunks << merge_name
+end
+puts merged_chunks.inspect
+
+# unlink all chunks
+chunks.each do |chunk|
+  File.unlink(chunk)
 end
-exit
 
-chunks.each do|wave|
+merged_chunks.each do|wave|
   chunk_file = wave.gsub(/\.wav/,'') + ".wav"
   system("mv #{wave} #{chunk_file}")
   system("speech2text #{chunk_file}")

merge-wave.c

@@ -38,7 +38,6 @@ int main(int argc, char **argv) {
   int *buffer = NULL;
   int i = 0;
 
-
   if (argc < 4) {
     fprintf(stderr, "usage: %s a1.wav a2.wav ... out.wav\n", argv[0]);
     return 1;
@@ -47,17 +46,17 @@ int main(int argc, char **argv) {
   memset(&outinfo, 0, sizeof(SF_INFO));
 
   for (i = 1; i < argc-1; ++i ) {
-    printf("scan: %s\n", argv[i]);
+    //printf("scan: %s\n", argv[i]);
     if (update_output_info(argv[i], &outinfo)) {
       return 1;
     }
   }
 
-  printf("%s will be %.2f seconds\n", argv[3], ((double)outinfo.frames / (double)outinfo.samplerate));
+  printf("%s will be %.2f seconds\n", argv[argc-1], ((double)outinfo.frames / (double)outinfo.samplerate));
 
-  out = sf_open(argv[3], SFM_WRITE, &outinfo);
+  out = sf_open(argv[argc-1], SFM_WRITE, &outinfo);
   if (!out) {
-    fprintf(stderr, "unable to write file: %s\n", argv[3]);
+    fprintf(stderr, "unable to write file: %s\n", argv[argc-1]);
     return 3;
   }
 

merge.rb

@@ -6,7 +6,7 @@ class Dur < Speech::AudioInspector::Duration
 accom = Dur.new("00:00:00.00")
 
 merge_set = []
-
+# decades
 ["00:00:00.32",
 "00:00:04.73",
 "00:00:01.82",