[Bugfix] Relax lang pin for voxtral (#21833)

Signed-off-by: Sanchit Gandhi <sgandhi3141@gmail.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>

vllm/entrypoints/openai/speech_to_text.py

@@ -86,11 +86,7 @@ class OpenAISpeechToText(OpenAIServing):
         audio_data: bytes,
     ) -> tuple[list[PromptType], float]:
         # Validate request
-        # TODO language should be optional and can be guessed.
+        language = self.model_cls.validate_language(request.language)
-        # For now we default to en. See
-        # https://github.com/huggingface/transformers/blob/main/src/transformers/models/whisper/generation_whisper.py#L1520
-        lang = request.language or "en"
-        self.model_cls.validate_language(lang)
 
         if len(audio_data) / 1024**2 > self.max_audio_filesize_mb:
             raise ValueError("Maximum file size exceeded.")
@@ -112,7 +108,7 @@ class OpenAISpeechToText(OpenAIServing):
                 audio=chunk,
                 stt_config=self.asr_config,
                 model_config=self.model_config,
-                language=lang,
+                language=language,
                 task_type=self.task_type,
                 request_prompt=request.prompt)
             prompts.append(prompt)

vllm/model_executor/models/interfaces.py

@@ -1,13 +1,14 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
-from collections.abc import Iterable, MutableSequence
+from collections.abc import Iterable, Mapping, MutableSequence
 from typing import (TYPE_CHECKING, ClassVar, Literal, Optional, Protocol,
                     Union, overload, runtime_checkable)
 
 import numpy as np
 import torch
 from torch import Tensor
+from transformers.models.whisper.tokenization_whisper import LANGUAGES
 from typing_extensions import Self, TypeIs
 
 from vllm.config import ModelConfig, SpeechToTextConfig
@@ -685,6 +686,8 @@ class SupportsQuant:
 @runtime_checkable
 class SupportsTranscription(Protocol):
     """The interface required for all models that support transcription."""
+    # Mapping from ISO639_1 language codes: language names
+    supported_languages: ClassVar[Mapping[str, str]]
 
     supports_transcription: ClassVar[Literal[True]] = True
 
@@ -694,11 +697,22 @@ class SupportsTranscription(Protocol):
     `True`.
     """
 
+    def __init_subclass__(cls, **kwargs):
+        super().__init_subclass__(**kwargs)
+        # language codes in supported_languages
+        # that don't exist in the full language map
+        invalid = set(cls.supported_languages) - set(LANGUAGES.keys())
+        if invalid:
+            raise ValueError(
+                f"{cls.__name__}.supported_languages contains invalid "
+                f"language codes: {sorted(invalid)}\n. "
+                f"Valid choices are: {sorted(LANGUAGES.keys())}")
+
     @classmethod
     def get_generation_prompt(cls, audio: np.ndarray,
                               stt_config: SpeechToTextConfig,
-                              model_config: ModelConfig, language: str,
+                              model_config: ModelConfig,
-                              task_type: str,
+                              language: Optional[str], task_type: str,
                               request_prompt: str) -> PromptType:
         """Get the prompt for the ASR model.
         The model has control over the construction, as long as it
@@ -706,9 +720,36 @@ class SupportsTranscription(Protocol):
         ...
 
     @classmethod
-    def validate_language(cls, language: str) -> bool:
+    def get_other_languages(cls) -> Mapping[str, str]:
-        """Check if the model supports a specific ISO639_1 language."""
+        # other possible language codes from the whisper map
-        ...
+        return {
+            k: v
+            for k, v in LANGUAGES.items() if k not in cls.supported_languages
+        }
+
+    @classmethod
+    def validate_language(cls, language: Optional[str]) -> Optional[str]:
+        """
+        Ensure the language specified in the transcription request 
+        is a valid ISO 639-1 language code. If the request language is 
+        valid, but not natively supported by the model, trigger a 
+        warning (but not an exception).
+        """
+        if language is None or language in cls.supported_languages:
+            return language
+        elif language in cls.get_other_languages():
+            logger.warning(
+                "Language %r is not natively supported by %s; "
+                "results may be less accurate. Supported languages: %r",
+                language,
+                cls.__name__,
+                list(cls.supported_languages.keys()),
+            )
+            return language
+        else:
+            raise ValueError(
+                f"Unsupported language: {language!r}.  Must be one of "
+                f"{list(cls.supported_languages.keys())}.")
 
     @classmethod
     def get_speech_to_text_config(

vllm/model_executor/models/voxtral.py

@@ -26,8 +26,7 @@ from vllm.logger import init_logger
 from vllm.model_executor.model_loader.weight_utils import default_weight_loader
 from vllm.model_executor.models import SupportsPP
 # yapf: disable
-from vllm.model_executor.models.whisper import (
+from vllm.model_executor.models.whisper import WhisperEncoder
-    WhisperEncoder, WhisperForConditionalGeneration)
 # yapf: enable
 from vllm.model_executor.sampling_metadata import SamplingMetadata
 from vllm.multimodal import MULTIMODAL_REGISTRY
@@ -50,6 +49,18 @@ from .utils import (flatten_bn, init_vllm_registered_model, maybe_prefix,
 
 logger = init_logger(__name__)
 
+ISO639_1_SUPPORTED_LANGS = {
+    "ar": "Arabic",
+    "nl": "Dutch",
+    "en": "English",
+    "fr": "French",
+    "de": "German",
+    "hi": "Hindi",
+    "it": "Italian",
+    "pt": "Portuguese",
+    "es": "Spanish",
+}
+
 
 class VoxtralProcessorAdapter:
     """
@@ -301,6 +312,7 @@ class VoxtralMultiModalProcessor(BaseMultiModalProcessor[VoxtralProcessingInfo]
                                         dummy_inputs=VoxtralDummyInputsBuilder)
 class VoxtralForConditionalGeneration(nn.Module, SupportsMultiModal,
                                       SupportsPP, SupportsTranscription):
+    supported_languages = ISO639_1_SUPPORTED_LANGS
 
     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
@@ -441,8 +453,8 @@ class VoxtralForConditionalGeneration(nn.Module, SupportsMultiModal,
     # for speech-to-text transcription
     def get_generation_prompt(cls, audio: np.ndarray,
                               model_config: ModelConfig,
-                              stt_config: SpeechToTextConfig, language: str,
+                              stt_config: SpeechToTextConfig,
-                              task_type: str,
+                              language: Optional[str], task_type: str,
                               request_prompt: str) -> PromptType:
         tokenizer = cached_tokenizer_from_config(model_config)
         audio = Audio(audio, int(stt_config.sample_rate),
@@ -457,11 +469,6 @@ class VoxtralForConditionalGeneration(nn.Module, SupportsMultiModal,
         prompts_dict["prompt_token_ids"] = tokenized.tokens
         return cast(PromptType, prompts_dict)
 
-    @classmethod
-    def validate_language(cls, language: str) -> bool:
-        # same as whisper
-        return WhisperForConditionalGeneration.validate_language(language)
-
     @classmethod
     def get_num_audio_tokens(cls, audio_duration_s: float,
                              stt_config: SpeechToTextConfig,

vllm/model_executor/models/whisper.py

@@ -109,51 +109,6 @@ ISO639_1_SUPPORTED_LANGS = {
     "vi": "Vietnamese",
     "cy": "Welsh"
 }
-ISO639_1_OTHER_LANGS = {
-    "lo": "Lao",
-    "jw": "Javanese",
-    "tk": "Turkmen",
-    "yi": "Yiddish",
-    "so": "Somali",
-    "bn": "Bengali",
-    "nn": "Norwegian Nynorsk",
-    "si": "Sinhala",
-    "yo": "Yoruba",
-    "sa": "Sanskrit",
-    "mi": "Māori",
-    "fo": "Faroese",  # codespell:ignore
-    "mt": "Maltese",
-    "tg": "Tajik",
-    "mg": "Malagasy",
-    "haw": "Hawaiian",
-    "km": "Khmer",
-    "br": "Breton",
-    "ps": "Pashto",
-    "ln": "Lingala",
-    "la": "Latin",
-    "ml": "Malayalam",
-    "sq": "Albanian",
-    "su": "Sundanese",
-    "eu": "Basque",
-    "ka": "Georgian",
-    "uz": "Uzbek",
-    "sn": "Shona",
-    "ht": "Haitian",
-    "as": "Assamese",
-    "mn": "Mongolian",
-    "te": "Telugu",
-    "pa": "Panjabi",
-    "tt": "Tatar",
-    "gu": "Gujarati",
-    "oc": "Occitan",
-    "ha": "Hausa",
-    "ba": "Bashkir",
-    "my": "Burmese",
-    "sd": "Sindhi",
-    "am": "Amharic",
-    "lb": "Luxembourgish",
-    "bo": "Tibetan"
-}
 
 
 class WhisperAudioInputs(TypedDict):
@@ -807,22 +762,20 @@ class WhisperForConditionalGeneration(nn.Module, SupportsTranscription,
 
     # Whisper only supports audio-conditioned generation.
     supports_transcription_only = True
+    supported_languages = ISO639_1_SUPPORTED_LANGS
 
     @classmethod
-    def validate_language(cls, language: str) -> bool:
+    def validate_language(cls, language: Optional[str]) -> Optional[str]:
-        if language in ISO639_1_SUPPORTED_LANGS:
+        if language is None:
-            return True
+            # TODO language should be optional and can be guessed.
-        elif language in ISO639_1_OTHER_LANGS:
+            # For now we default to en. See
+            # https://github.com/huggingface/transformers/blob/main/src/transformers/models/whisper/generation_whisper.py#L1520
             logger.warning(
-                "The selected language %s has limited accuracy with"
+                "Defaulting to language='en'. If you wish to transcribe "
-                " reported WER>=0.5. Results may be less accurate "
+                "audio in a different language, pass the `language` field "
-                "for this choice.", language)
+                "in the TranscriptionRequest.")
-            return True
+            language = "en"
-        else:
+        return super().validate_language(language)
-            raise ValueError(f"Unsupported language: {language}."
-                             "Language should be one of:" +
-                             f" {list(ISO639_1_SUPPORTED_LANGS.values())}" +
-                             f"or {list(ISO639_1_OTHER_LANGS.values())}")
 
     @classmethod
     def get_generation_prompt(
@@ -830,9 +783,12 @@ class WhisperForConditionalGeneration(nn.Module, SupportsTranscription,
             audio: np.ndarray,
             model_config: ModelConfig,  # not needed here
             stt_config: SpeechToTextConfig,
-            language: str,
+            language: Optional[str],
             task_type: str,
             request_prompt: str) -> PromptType:
+        if language is None:
+            raise ValueError(
+                "Language must be specified when creating the Whisper prompt")
         prompt = {
             "encoder_prompt": {
                 # Whisper does not support encoder prompt.