Skip to content

feat(speech): add Vulkan whisper-tiny variants #1398

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
msluszniak wants to merge 1 commit into from
Closed

feat(speech): add Vulkan whisper-tiny variants #1398

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions apps/speech/app/audio-file-transcription/index.tsx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,6 +73,16 @@ const MODELS = [
config: models.speechToText.WHISPER.TINY.MLX_BF16,
disabled: Platform.OS !== 'ios',
},
{
name: 'Tiny Multilingual (Vulkan fp16)',
config: models.speechToText.WHISPER.TINY.VULKAN_FP16,
disabled: Platform.OS !== 'android',
},
{
name: 'Tiny Multilingual (Vulkan int8)',
config: models.speechToText.WHISPER.TINY.VULKAN_INT8,
disabled: Platform.OS !== 'android',
},
{
name: 'Base Multilingual (CPU)',
config: models.speechToText.WHISPER.BASE.XNNPACK_FP32,
Expand Down
10 changes: 10 additions & 0 deletions apps/speech/app/microphone-transcription/index.tsx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -66,6 +66,16 @@ const MODELS = [
config: models.speechToText.WHISPER.TINY.MLX_BF16,
disabled: Platform.OS !== 'ios',
},
{
name: 'Tiny Multilingual (Vulkan fp16)',
config: models.speechToText.WHISPER.TINY.VULKAN_FP16,
disabled: Platform.OS !== 'android',
},
{
name: 'Tiny Multilingual (Vulkan int8)',
config: models.speechToText.WHISPER.TINY.VULKAN_INT8,
disabled: Platform.OS !== 'android',
},
{
name: 'Base Multilingual (CPU)',
config: models.speechToText.WHISPER.BASE.XNNPACK_FP32,
Expand Down
4 changes: 2 additions & 2 deletions packages/react-native-executorch/scripts/download-libs.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,8 +124,8 @@ const FEATURE_MAP = {
multimodalLLM: { backends: ['xnnpack', 'mlx', 'vulkan'], libs: ['opencv'] },
// Privacy filter classifiers ship xnnpack + an MLX iOS export.
privacyFilter: { backends: ['xnnpack', 'mlx'], libs: [] },
// Whisper ships xnnpack + coreml.
speechToText: { backends: ['xnnpack', 'coreml'], libs: [] },
// Whisper ships xnnpack + coreml + a Vulkan Android export.
speechToText: { backends: ['xnnpack', 'coreml', 'vulkan'], libs: [] },
// Kokoro ships xnnpack only.
textToSpeech: { backends: ['xnnpack'], libs: ['phonemis'] },
// FSMN VAD — xnnpack only.
Expand Down
14 changes: 14 additions & 0 deletions packages/react-native-executorch/src/models.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -750,6 +750,18 @@ const WHISPER_TINY_MLX_INT8: WhisperSttModel = {
supportedLanguages: WHISPER_LANGUAGES,
vadModel: FSMN_VAD_XNNPACK_FP32,
};
const WHISPER_TINY_VULKAN_FP16: WhisperSttModel = {
modelPath: `${BASE_URL}-whisper-tiny/${NEXT_VERSION_TAG}/vulkan/whisper_tiny_vulkan_fp16.pte`,
tokenizerPath: `${BASE_URL}-whisper-tiny/${NEXT_VERSION_TAG}/tokenizer.json`,
supportedLanguages: WHISPER_LANGUAGES,
vadModel: FSMN_VAD_XNNPACK_FP32,
};
const WHISPER_TINY_VULKAN_INT8: WhisperSttModel = {
modelPath: `${BASE_URL}-whisper-tiny/${NEXT_VERSION_TAG}/vulkan/whisper_tiny_vulkan_int8.pte`,
tokenizerPath: `${BASE_URL}-whisper-tiny/${NEXT_VERSION_TAG}/tokenizer.json`,
supportedLanguages: WHISPER_LANGUAGES,
vadModel: FSMN_VAD_XNNPACK_FP32,
};

const WHISPER_BASE_EN_XNNPACK_FP32: WhisperSttModel<'en'> = {
modelPath: `${BASE_URL}-whisper-base.en/${NEXT_VERSION_TAG}/xnnpack/whisper_base_en_xnnpack_fp32.pte`,
Expand Down Expand Up @@ -1862,6 +1874,8 @@ export const models = {
COREML_FP16: WHISPER_TINY_COREML_FP16,
MLX_BF16: WHISPER_TINY_MLX_BF16,
MLX_INT8: WHISPER_TINY_MLX_INT8,
VULKAN_FP16: WHISPER_TINY_VULKAN_FP16,
VULKAN_INT8: WHISPER_TINY_VULKAN_INT8,
},
/**
* Multilingual Whisper Base model. Higher accuracy across supported
Expand Down