Why does in-code streaming sound different (flatter)?

[jhfoo]

Tried the examples:

1. Pipe raw piper output from shell to aplay:

echo 'This sentence is spoken first. This sentence is synthesized while the first sentence is spoken.' | \
  ./piper --model en_US-lessac-medium.onnx --output-raw | \
  aplay -r 22050 -f S16_LE -t raw -

2. Stream from code (python):

dtype = 'int16'
DeviceId = 3
stream = sd.OutputStream(
  samplerate=voice.config.sample_rate, 
  channels=1, 
  device = DeviceId, 
  dtype=dtype
)
stream.start()
for bytes in voice.synthesize_stream_raw(text):
    data = np.frombuffer(bytes, dtype=np.int16)
    stream.write(data)

Option 2 works but the sound is not as bold as Option 1. Plus the first word is not enunciated as clearly. Both output to the same speaker device.

What might be causing this?