Transcribe audio files with Streaming

1
import assemblyai as aai
2
from assemblyai.streaming.v3 import (
3
    BeginEvent,
4
    StreamingClient,
5
    StreamingClientOptions,
6
    StreamingError,
7
    StreamingEvents,
8
    StreamingParameters,
9
    TerminationEvent,
10
    TurnEvent
11
)
12
from typing import Type
13
14
def on_begin(self: Type[StreamingClient], event: BeginEvent):
15
  "This function is called when the connection has been established."
16
17
  print("Session ID:", event.id)
18
19
def on_turn(self: Type[StreamingClient], event: TurnEvent):
20
  "This function is called when a new transcript has been received."
21
22
  print(event.transcript, end="\r\n")
23
24
def on_terminated(self: Type[StreamingClient], event: TerminationEvent):
25
  "This function is called when an error occurs."
26
27
  print(
28
    f"Session terminated: {event.audio_duration_seconds} seconds of audio processed"
29
  )
30
31
def on_error(self: Type[StreamingClient], error: StreamingError):
32
  "This function is called when the connection has been closed."
33
34
  print(f"Error occurred: {error}")
35
36
37
# Create the streaming client
38
client = StreamingClient(
39
  StreamingClientOptions(
40
    api_key="YOUR_API_KEY"
41
  )
42
)
43
44
client.on(StreamingEvents.Begin, on_begin)
45
client.on(StreamingEvents.Turn, on_turn)
46
client.on(StreamingEvents.Termination, on_terminated)
47
client.on(StreamingEvents.Error, on_error)
48
49
def stream_file(filepath: str, sample_rate: int):
50
    """Stream audio file in 50ms chunks instead of 300ms"""
51
    import time
52
    import wave
53
54
    chunk_duration = 0.1
55
56
    with wave.open(filepath, 'rb') as wav_file:
57
        if wav_file.getnchannels() != 1:
58
            raise ValueError("Only mono audio is supported")
59
60
        file_sample_rate = wav_file.getframerate()
61
        if file_sample_rate != sample_rate:
62
            print(f"Warning: File sample rate ({file_sample_rate}) doesn't match expected rate ({sample_rate})")
63
64
        frames_per_chunk = int(file_sample_rate * chunk_duration)
65
66
        while True:
67
            frames = wav_file.readframes(frames_per_chunk)
68
69
            if not frames:
70
                break
71
72
            yield frames
73
74
            time.sleep(chunk_duration)
75
76
file_stream = stream_file(
77
  filepath="audio.wav",
78
  sample_rate=44100,
79
)
80
81
client.connect(
82
    StreamingParameters(
83
        sample_rate=48000,
84
        format_turns=True,
85
    )
86
)
87
88
try:
89
    client.stream(file_stream)
90
finally:
91
    client.disconnect(terminate=True)

1
import assemblyai as aai
2
from assemblyai.streaming.v3 import (
3
    BeginEvent,
4
    StreamingClient,
5
    StreamingClientOptions,
6
    StreamingError,
7
    StreamingEvents,
8
    StreamingParameters,
9
    TerminationEvent,
10
    TurnEvent
11
)
12
from typing import Type

1
def on_begin(self: Type[StreamingClient], event: BeginEvent):
2
  "This function is called when the connection has been established."
3
4
  print("Session ID:", event.id)
5
6
def on_turn(self: Type[StreamingClient], event: TurnEvent):
7
  "This function is called when a new transcript has been received."
8
9
  print(event.transcript, end="\r\n")
10
11
def on_terminated(self: Type[StreamingClient], event: TerminationEvent):
12
  "This function is called when an error occurs."
13
14
  print(
15
    f"Session terminated: {event.audio_duration_seconds} seconds of audio processed"
16
  )
17
18
def on_error(self: Type[StreamingClient], error: StreamingError):
19
  "This function is called when the connection has been closed."
20
21
  print(f"Error occurred: {error}")

1
# Create the streaming client
2
client = StreamingClient(
3
  StreamingClientOptions(
4
    api_key="YOUR_API_KEY"
5
  )
6
)
7
8
client.on(StreamingEvents.Begin, on_begin)
9
client.on(StreamingEvents.Turn, on_turn)
10
client.on(StreamingEvents.Termination, on_terminated)
11
client.on(StreamingEvents.Error, on_error)

1
def stream_file(filepath: str, sample_rate: int):
2
    """Stream audio file in 50ms chunks instead of 300ms"""
3
    import time
4
    import wave
5
6
    chunk_duration = 0.1
7
8
    with wave.open(filepath, 'rb') as wav_file:
9
        if wav_file.getnchannels() != 1:
10
            raise ValueError("Only mono audio is supported")
11
12
        file_sample_rate = wav_file.getframerate()
13
        if file_sample_rate != sample_rate:
14
            print(f"Warning: File sample rate ({file_sample_rate}) doesn't match expected rate ({sample_rate})")
15
16
        frames_per_chunk = int(file_sample_rate * chunk_duration)
17
18
        while True:
19
            frames = wav_file.readframes(frames_per_chunk)
20
21
            if not frames:
22
                break
23
24
            yield frames
25
26
            time.sleep(chunk_duration)
27
28
file_stream = stream_file(
29
  filepath="audio.wav",
30
  sample_rate=44100,
31
)

1
client.connect(
2
    StreamingParameters(
3
        sample_rate=48000,
4
        format_turns=True,
5
    )
6
)

1
try:
2
    client.stream(file_stream)
3
finally:
4
    client.disconnect(terminate=True)