stevenhillis commited on
Commit
ccbb9b9
1 Parent(s): 6a9e916

send encoded bytes not numpy

Browse files
Files changed (1) hide show
  1. app.py +8 -2
app.py CHANGED
@@ -10,10 +10,16 @@ base_url = "https://api.sandbox.deepgram.com/nlu"
10
  token_str = os.environ['DG_TOKEN']
11
  def tts_fn(text, prompt_audio, prompt_seconds, inference_steps, inference_temperature, pitch_steps):
12
  texts = [text]
13
- prompt_audio = np.reshape(prompt_audio[1], (1, 1, -1)).astype(np.float32, order='C') / 32768.0
 
 
 
 
 
 
14
  response = requests.post(
15
  f'{base_url}',
16
- files=[('texts', ('texts', json.dumps(texts), 'application/json')), ('prompt_audio', ('prompt_audio', json.dumps(prompt_audio.tolist()), 'application/json'))],
17
  params={'synthesize': 'true', 'pitch_steps': int(pitch_steps), 'soundstorm_steps': inference_steps, 'temperature': inference_temperature, 'prompt_seconds': prompt_seconds},
18
  headers={
19
  'Authorization': f'Token {token_str}'
 
10
  token_str = os.environ['DG_TOKEN']
11
  def tts_fn(text, prompt_audio, prompt_seconds, inference_steps, inference_temperature, pitch_steps):
12
  texts = [text]
13
+ sr = prompt_audio[0]
14
+ prompt_audio = np.reshape(prompt_audio[1], (1, -1)).astype(np.float32, order='C') / 32768.0
15
+ audio_bytes = bytes()
16
+ byte_io = io.BytesIO(audio_bytes)
17
+ wavfile.write(byte_io, sr, prompt_audio)
18
+ prompt_audio = [base64.b64encode(byte_io).decode('utf-8')]
19
+
20
  response = requests.post(
21
  f'{base_url}',
22
+ files=[('texts', ('texts', json.dumps(texts), 'application/json')), ('prompt_audio', ('prompt_audio', json.dumps(prompt_audio), 'application/json'))],
23
  params={'synthesize': 'true', 'pitch_steps': int(pitch_steps), 'soundstorm_steps': inference_steps, 'temperature': inference_temperature, 'prompt_seconds': prompt_seconds},
24
  headers={
25
  'Authorization': f'Token {token_str}'