[Frontend][Docs] Transcription API streaming (#13301)

Signed-off-by: NickLucche <nlucches@redhat.com>
This commit is contained in:
Nicolò Lucchesi
2025-03-06 11:39:35 +01:00
committed by GitHub
parent 69ff99fdcd
commit fa82b93853
5 changed files with 297 additions and 26 deletions

View File

@@ -1,4 +1,8 @@
# SPDX-License-Identifier: Apache-2.0
import asyncio
import json
import httpx
from openai import OpenAI
from vllm.assets.audio import AudioAsset
@@ -13,11 +17,50 @@ client = OpenAI(
api_key=openai_api_key,
base_url=openai_api_base,
)
with open(str(mary_had_lamb), "rb") as f:
transcription = client.audio.transcriptions.create(
file=f,
model="openai/whisper-large-v3",
language="en",
response_format="text",
temperature=0.0)
print("transcription result:", transcription)
def sync_openai():
with open(str(mary_had_lamb), "rb") as f:
transcription = client.audio.transcriptions.create(
file=f,
model="openai/whisper-small",
language="en",
response_format="json",
temperature=0.0)
print("transcription result:", transcription.text)
sync_openai()
# OpenAI Transcription API client does not support streaming.
async def stream_openai_response():
data = {
"language": "en",
'stream': True,
"model": "openai/whisper-large-v3",
}
url = openai_api_base + "/audio/transcriptions"
print("transcription result:", end=' ')
async with httpx.AsyncClient() as client:
with open(str(winning_call), "rb") as f:
async with client.stream('POST', url, files={'file': f},
data=data) as response:
async for line in response.aiter_lines():
# Each line is a JSON object prefixed with 'data: '
if line:
if line.startswith('data: '):
line = line[len('data: '):]
# Last chunk, stream ends
if line.strip() == '[DONE]':
break
# Parse the JSON response
chunk = json.loads(line)
# Extract and print the content
content = chunk['choices'][0].get('delta',
{}).get('content')
print(content, end='')
# Run the asynchronous function
asyncio.run(stream_openai_response())