From 65be443e982755a96915b363af6b6ad2dfe5c827 Mon Sep 17 00:00:00 2001
From: "jiaqi.sjq" <jiaqi.sjq@alibaba-inc.com>
Date: Tue, 11 Oct 2022 17:22:58 +0800
Subject: [PATCH] [to #41669377] Add more models to test in tts UT

Link https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10360754#tab=detail
        Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/10360754
---
 .../audio/tts/models/datasets/__init__.py     |  0
 tests/pipelines/test_text_to_speech.py        | 70 +++++++++++++++----
 2 files changed, 58 insertions(+), 12 deletions(-)
 mode change 100755 => 100644 modelscope/models/audio/tts/models/datasets/__init__.py

diff --git a/modelscope/models/audio/tts/models/datasets/__init__.py b/modelscope/models/audio/tts/models/datasets/__init__.py
old mode 100755
new mode 100644
diff --git a/tests/pipelines/test_text_to_speech.py b/tests/pipelines/test_text_to_speech.py
index f659e59b..0caf1c84 100644
--- a/tests/pipelines/test_text_to_speech.py
+++ b/tests/pipelines/test_text_to_speech.py
@@ -27,21 +27,67 @@ class TextToSpeechSambertHifigan16kPipelineTest(unittest.TestCase,
 
     def setUp(self) -> None:
         self.task = Tasks.text_to_speech
-        self.model_id = 'damo/speech_sambert-hifigan_tts_zhitian_emo_zh-cn_16k'
+        zhcn_text = '今天北京天气怎么样'
+        en_text = 'How is the weather in Beijing?'
+        zhcn_voice = ['zhitian_emo', 'zhizhe_emo', 'zhiyan_emo', 'zhibei_emo']
+        enus_voice = ['andy', 'annie']
+        engb_voice = ['luca', 'luna']
+        self.tts_test_cases = []
+        for voice in zhcn_voice:
+            model_id = 'damo/speech_sambert-hifigan_tts_%s_%s_16k' % (voice,
+                                                                      'zh-cn')
+            self.tts_test_cases.append({
+                'voice': voice,
+                'model_id': model_id,
+                'text': zhcn_text
+            })
+        for voice in enus_voice:
+            model_id = 'damo/speech_sambert-hifigan_tts_%s_%s_16k' % (voice,
+                                                                      'en-us')
+            self.tts_test_cases.append({
+                'voice': voice,
+                'model_id': model_id,
+                'text': en_text
+            })
+        for voice in engb_voice:
+            model_id = 'damo/speech_sambert-hifigan_tts_%s_%s_16k' % (voice,
+                                                                      'en-gb')
+            self.tts_test_cases.append({
+                'voice': voice,
+                'model_id': model_id,
+                'text': en_text
+            })
+        zhcn_model_id = 'damo/speech_sambert-hifigan_tts_zh-cn_16k'
+        enus_model_id = 'damo/speech_sambert-hifigan_tts_en-us_16k'
+        engb_model_id = 'damo/speech_sambert-hifigan_tts_en-gb_16k'
+        self.tts_test_cases.append({
+            'voice': 'zhcn',
+            'model_id': zhcn_model_id,
+            'text': zhcn_text
+        })
+        self.tts_test_cases.append({
+            'voice': 'enus',
+            'model_id': enus_model_id,
+            'text': en_text
+        })
+        self.tts_test_cases.append({
+            'voice': 'engb',
+            'model_id': engb_model_id,
+            'text': en_text
+        })
 
     @unittest.skipUnless(test_level() >= 0, 'skip test in current test level')
     def test_pipeline(self):
-        text = '今天北京天气怎么样？'
-        voice = 'zhitian_emo'
-
-        model = Model.from_pretrained(
-            model_name_or_path=self.model_id, revision='pytorch_am')
-        sambert_hifigan_tts = pipeline(task=self.task, model=model)
-        self.assertTrue(sambert_hifigan_tts is not None)
-        output = sambert_hifigan_tts(input=text, voice=voice)
-        self.assertIsNotNone(output[OutputKeys.OUTPUT_PCM])
-        pcm = output[OutputKeys.OUTPUT_PCM]
-        write('output.wav', 16000, pcm)
+        for case in self.tts_test_cases:
+            logger.info('test %s' % case['voice'])
+            model = Model.from_pretrained(
+                model_name_or_path=case['model_id'], revision='pytorch_am')
+            sambert_hifigan_tts = pipeline(task=self.task, model=model)
+            self.assertTrue(sambert_hifigan_tts is not None)
+            output = sambert_hifigan_tts(input=case['text'])
+            self.assertIsNotNone(output[OutputKeys.OUTPUT_PCM])
+            pcm = output[OutputKeys.OUTPUT_PCM]
+            write('output_%s.wav' % case['voice'], 16000, pcm)
 
     @unittest.skip('demo compatibility test is only enabled on a needed-basis')
     def test_demo_compatibility(self):