Fix encoding g722 format

g722 format only supports 16k Hz, but AVCodec does not list this.
The implementation does not insert resampling and the resulting
audio can be slowed down or sped up.

Author: mthrok

test/torchaudio_unittest/io/stream_writer_test.py

@@ -389,6 +389,22 @@ def test_audio_num_frames_lossy(self, ext, num_channels, sample_rate):
             return
         self.assertEqual(saved.shape, data.shape)
 
+    def test_g722_sample_rate(self):
+        """Encoding G.722 properly converts sample rate to 16k"""
+        filename = "test.g722"
+        sample_rate = 41000
+        data = get_sinusoid(sample_rate=sample_rate, n_channels=1, channels_first=False)
+
+        # write data
+        dst = self.get_temp_path(filename)
+        w = StreamWriter(dst, format="g722")
+        w.add_audio_stream(sample_rate=sample_rate, num_channels=1)
+        with w.open():
+            w.write_audio_chunk(0, data)
+
+        r = StreamReader(src=self.get_temp_path(filename))
+        self.assertEqual(r.get_src_stream_info(0).sample_rate, 16000)
+
     def test_preserve_fps(self):
         """Decimal point frame rate is properly saved
 

torchaudio/csrc/ffmpeg/stream_writer/encode_process.cpp

@@ -273,6 +273,20 @@ int get_enc_sr(
     int src_sample_rate,
     const c10::optional<int>& encoder_sample_rate,
     const AVCodec* codec) {
+  // G.722 only supports 16000 Hz, but it does not list the sample rate in
+  // supported_samplerates so we hard code it here.
+  if (codec->id == AV_CODEC_ID_ADPCM_G722) {
+    if (encoder_sample_rate) {
+      auto val = encoder_sample_rate.value();
+      TORCH_CHECK(
+          val == 16'000,
+          codec->name,
+          " does not support sample rate ",
+          val,
+          ". Supported values are; 16000.");
+    }
+    return 16'000;
+  }
   if (encoder_sample_rate) {
     const int& encoder_sr = encoder_sample_rate.value();
     TORCH_CHECK(