MindIE部署Qwen2-audio模型怎么调用？_MindIE_昇腾论坛

# MindIE部署Qwen2-audio模型怎么调用？_MindIE_昇腾论坛

## 概述

本文档基于昇腾社区论坛帖子生成的技术教程。

**原始链接**: https://www.hiascend.com/forum/thread-0279188635892613220-1-1.html

**生成时间**: 2025-08-27 10:33:58

---

## 问题描述

mindie部署了Qwen2-audio但是尝试了很多endpoint都调不通请问是不是mindie没有开放audio模型的端口还是我调用的方式不对 import openai client = openai.OpenAI( base_url="http://x.x.x.x:21152/v1", api_key="none", ) audio_file = open("en.mp3", "rb") response = client.audio.transcriptions.create( model="qwen2-audio", # 你的模型ID file=audio_file, response_format="text", # 也可以是 "json""srt" 等 temperature=0, language="en" # 可选如果明确知道语种 ) print(response.text) # 这是转写文本 curl -X POST http://x.x.x.x:21152/v1/audio/transcriptions -H "Content-Type: multipart/form-data" -H "Authorization: Bearer none" -F "model=qwen2_audio" -F "file=@en.mp3" -F "response_format=text" -F "temperature=0" from openai import OpenAI base_url = "http://x.x.x.x:21152/v1" client = OpenAI( api_key="EMPTY", # 没有鉴权直接写 EMPTY 或 None base_url=base_url, ) # Chat Completions多模态文本+音频 chat_resp = client.chat.completions.create( model="qwen2_audio", messages=[ { "role": "user", "content": [ {"type": "text", "text": "请分析下这段音频的内容"}, # audio_type, 支持 audio_url {"type": "audio_url", "audio...

## 完整内容

mindie部署了Qwen2-audio但是尝试了很多endpoint都调不通请问是不是mindie没有开放audio模型的端口还是我调用的方式不对 import openai client = openai.OpenAI( base_url="http://x.x.x.x:21152/v1", api_key="none", ) audio_file = open("en.mp3", "rb") response = client.audio.transcriptions.create( model="qwen2-audio", # 你的模型ID file=audio_file, response_format="text", # 也可以是 "json""srt" 等 temperature=0, language="en" # 可选如果明确知道语种 ) print(response.text) # 这是转写文本 curl -X POST http://x.x.x.x:21152/v1/audio/transcriptions -H "Content-Type: multipart/form-data" -H "Authorization: Bearer none" -F "model=qwen2_audio" -F "file=@en.mp3" -F "response_format=text" -F "temperature=0" from openai import OpenAI base_url = "http://x.x.x.x:21152/v1" client = OpenAI( api_key="EMPTY", # 没有鉴权直接写 EMPTY 或 None base_url=base_url, ) # Chat Completions多模态文本+音频 chat_resp = client.chat.completions.create( model="qwen2_audio", messages=[ { "role": "user", "content": [ {"type": "text", "text": "请分析下这段音频的内容"}, # audio_type, 支持 audio_url {"type": "audio_url", "audio_url": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen2-Audio/audio/guess_age_gender.wav"} ] } ] ) print(chat_resp) 这是我按照华为mindie service重新搞的请求https://www.hiascend.com/document/detail/zh/mindie/20RC1/mindieservice/servicedev/mindie_service0319.html#ZH-CN_TOPIC_0000002252950588__zh-cn_topic_0000001909624241_zh-cn_topic_0000001821934202_zh-cn_topic_0000001799506204_section18279228143017但是返回值没有解释我的音频 (qwen_audio) root@bms-58150179-001:/data/model_data/llm_test# python test_llm.py ChatCompletion(id='endpoint_common_7', choices=[Choice(finish_reason='stop', index=0, logprobs=None, message=ChatCompletionMessage(content='', refusal=None, role='assistant', annotations=None, audio=None, function_call=None, tool_calls=None))], created=1753256000, model='qwen2_audio', object='chat.completion', service_tier=None, system_fingerprint=None, usage=CompletionUsage(completion_tokens=1, prompt_tokens=233, total_tokens=234, completion_tokens_details=None, prompt_tokens_details=None), prefill_time=106, decode_time_arr=[])

---

## 技术要点总结

基于以上内容，主要技术要点包括：

1. **问题类型**: 部署安装
2. **涉及技术**: 昇腾, HTTPS, AI, MindIE
3. **解决方案**: 请参考完整内容中的解决方案

## 相关资源

- 昇腾社区: https://www.hiascend.com/
- 昇腾论坛: https://www.hiascend.com/forum/

---

*本文档由AI自动生成，仅供参考。如有疑问，请参考原始帖子。*