Conversational speech tool, simualtor + unit tests

webrtc/modules/audio_processing/test/conversational_speech/simulator.h

+/*
+ *  Copyright (c) 2017 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef WEBRTC_MODULES_AUDIO_PROCESSING_TEST_CONVERSATIONAL_SPEECH_SIMULATOR_H_
+#define WEBRTC_MODULES_AUDIO_PROCESSING_TEST_CONVERSATIONAL_SPEECH_SIMULATOR_H_
+
+#include <map>
+#include <memory>
+#include <string>
+#include <utility>
+
+#include "webrtc/base/constructormagic.h"
+#include "webrtc/modules/audio_processing/test/conversational_speech/multiend_call.h"
+
+namespace webrtc {
+namespace test {
+namespace conversational_speech {
+
+struct SpeakerOutputFilePaths {
+  SpeakerOutputFilePaths(std::string new_near_end,
+                         std::string new_far_end)
+      : near_end(std::move(new_near_end)),

[minyue-webrtc, 2017/05/16 15:05:17]

I am not a fan of move string, is there a benefit?

[AleBzk, 2017/05/17 12:49:38]

Different reviewer, different tastes :)

I had a discussion exactly about this some time ago.
Karl and Henrik concluded that this is a good way to go.

Check at line 38 in
https://codereview.webrtc.org/2781573002/diff/130001/webrtc/modules/audio_pro...

[kwiberg-webrtc, 2017/05/18 13:28:49]

It's more efficient (since you don't have to make a new heap allocation, copy
the contents, and free the old heap allocation). And as long as you make sure to
not use the moved-from string between the point of the std::move and when it
goes out of scope, nothing can go wrong.

(If you nevertheless go with not moving, pass the arguments by const& instead.)

[minyue-webrtc, 2017/05/18 13:35:13]

I am concerned about the definition, from the signature
   SpeakerOutputFilePaths(std::string new_near_end, std::string new_far_end)
I won't be able to tell that I will gain efficiency if I call it with
   SpeakerOutputFilePaths(std::move(str_1), std::move(str_2))

SpeakerOutputFilePaths(std::string&& new_near_end, std::string&& new_far_end)

and 

SpeakerOutputFilePaths(const std::string& new_near_end, const std::string&
new_far_end)

are more obvious to me.

We can have both :)

+        far_end(std::move(new_far_end)) {}
+  // Paths to the near-end and far-end audio track files.
+  const std::string near_end;
+  const std::string far_end;
+};
+
+// Generates the near-end and far-end audio track pairs for each speaker.
+std::unique_ptr<std::map<std::string, SpeakerOutputFilePaths>>
+    Simulate(const MultiEndCall& multiend_call, const std::string& output_path);
+
+}  // namespace conversational_speech
+}  // namespace test
+}  // namespace webrtc
+
+#endif  // WEBRTC_MODULES_AUDIO_PROCESSING_TEST_CONVERSATIONAL_SPEECH_SIMULATOR_H_