[BugFix] Use unique ids for different transcription prompts (#29372)

njhill · web-flow · commit 7992324f2347 · 2025-11-25T06:55:16.000Z
Signed-off-by: Nick Hill &lt;nhill@redhat.com&gt;
diff --git a/vllm/entrypoints/openai/speech_to_text.py b/vllm/entrypoints/openai/speech_to_text.py
@@ -201,10 +201,10 @@ async def _create_speech_to_text(
                 self.engine_client.generate(
                     prompt,
                     sampling_params,
-                    request_id,
+                    f"{request_id}_{i}",
                     lora_request=lora_request,
                 )
-                for prompt in prompts
+                for i, prompt in enumerate(prompts)
             ]
         except ValueError as e:
             # TODO: Use a vllm-specific Validation Error

Original file line number	Diff line number	Diff line change
`@@ -201,10 +201,10 @@ async def _create_speech_to_text(`
`201`	`201`	`self.engine_client.generate(`
`202`	`202`	`prompt,`
`203`	`203`	`sampling_params,`
`204`		`- request_id,`
	`204`	`+ f"{request_id}_{i}",`
`205`	`205`	`lora_request=lora_request,`
`206`	`206`	`)`
`207`		`- for prompt in prompts`
	`207`	`+ for i, prompt in enumerate(prompts)`
`208`	`208`	`]`
`209`	`209`	`except ValueError as e:`
`210`	`210`	`# TODO: Use a vllm-specific Validation Error`