Batch Speech-to-Text (STT) API Tutorial Using Saarika Model

1
!pip install -U sarvamai

1
API_KEY = "YOUR_API_KEY_HERE"

1
from pathlib import Path
2
from sarvamai import SarvamAI
3
4
API_KEY = "YOUR_API_KEY_HERE"
5
audio_files = ["/path/to/your/audio1.mp3", "/path/to/your/audio2.mp3"]  # Update with your file paths
6
output_dir = Path("/output")
7
output_dir.mkdir(exist_ok=True)
8
9
def run_stt_sync():
10
    client = SarvamAI(api_subscription_key=API_KEY)
11
    
12
    # Create and configure batch STT job
13
    job = client.speech_to_text_job.create_job(
14
        model="saarika:v2.5",
15
        with_diarization=True,
16
        with_timestamps=True,
17
        language_code="en-IN",
18
        num_speakers=2,
19
    )
20
    print(f"Job created: {job._job_id}")
21
    
22
    # Upload and process files
23
    job.upload_files(file_paths=audio_files, timeout=120.0)
24
    job.start()
25
    print("Transcription started...")
26
    
27
    # Wait for completion
28
    job.wait_until_complete(poll_interval=5, timeout=600)
29
    
30
    # Check file-level results
31
    file_results = job.get_file_results()
32
    
33
    print(f"\nSuccessful: {len(file_results['successful'])}")
34
    for f in file_results['successful']:
35
        print(f"  ✓ {f['file_name']}")
36
    
37
    print(f"\nFailed: {len(file_results['failed'])}")
38
    for f in file_results['failed']:
39
        print(f"  ✗ {f['file_name']}: {f['error_message']}")
40
    
41
    # Handle all files failed
42
    if len(file_results['successful']) == 0:
43
        print("\nAll files failed.")
44
        return
45
    
46
    # Download outputs for successful files
47
    job.download_outputs(output_dir=str(output_dir))
48
    print(f"\nDownloaded {len(file_results['successful'])} file(s) to: {output_dir}")
49
50
run_stt_sync()

1
import asyncio
2
from pathlib import Path
3
from sarvamai import AsyncSarvamAI
4
5
API_KEY = "YOUR_API_KEY_HERE"
6
audio_files = ["/path/to/your/audio1.mp3", "/path/to/your/audio2.mp3"]  # Update with your file paths
7
output_dir = Path("/output")
8
output_dir.mkdir(exist_ok=True)
9
10
async def run_stt_async_job():
11
    client = AsyncSarvamAI(api_subscription_key=API_KEY)
12
    
13
    # Create and configure batch STT job
14
    job = await client.speech_to_text_job.create_job(
15
        model="saarika:v2.5",
16
        with_diarization=True,
17
        with_timestamps=True,
18
        language_code="en-IN",
19
        num_speakers=2,
20
    )
21
    print(f"Job created: {job._job_id}")
22
    
23
    # Upload and process files
24
    await job.upload_files(file_paths=audio_files, timeout=120.0)
25
    await job.start()
26
    print("Transcription started...")
27
    
28
    # Wait for completion
29
    await job.wait_until_complete(poll_interval=5, timeout=600)
30
    
31
    # Check file-level results
32
    file_results = await job.get_file_results()
33
    
34
    print(f"\nSuccessful: {len(file_results['successful'])}")
35
    for f in file_results['successful']:
36
        print(f"  ✓ {f['file_name']}")
37
    
38
    print(f"\nFailed: {len(file_results['failed'])}")
39
    for f in file_results['failed']:
40
        print(f"  ✗ {f['file_name']}: {f['error_message']}")
41
    
42
    # Handle all files failed
43
    if len(file_results['successful']) == 0:
44
        print("\nAll files failed.")
45
        return
46
    
47
    # Download outputs for successful files
48
    await job.download_outputs(output_dir=str(output_dir))
49
    print(f"\nDownloaded {len(file_results['successful'])} file(s) to: {output_dir}")
50
51
# For Jupyter environments:
52
import nest_asyncio
53
nest_asyncio.apply()
54
await run_stt_async_job()