Fix building (#1343)

This commit is contained in:
Fangjun Kuang
2024-09-13 13:33:52 +08:00
committed by GitHub
parent 65cfa7548a
commit 544857b097
17 changed files with 190 additions and 44 deletions

View File

@@ -155,6 +155,14 @@ function initSherpaOnnxOnlineModelConfig(config, Module) {
};
}
if (!('tokensBuf' in config)) {
config.tokensBuf = '';
}
if (!('tokensBufSize' in config)) {
config.tokensBufSize = 0;
}
const transducer =
initSherpaOnnxOnlineTransducerModelConfig(config.transducer, Module);
@@ -164,7 +172,7 @@ function initSherpaOnnxOnlineModelConfig(config, Module) {
const ctc = initSherpaOnnxOnlineZipformer2CtcModelConfig(
config.zipformer2Ctc, Module);
const len = transducer.len + paraformer.len + ctc.len + 7 * 4;
const len = transducer.len + paraformer.len + ctc.len + 9 * 4;
const ptr = Module._malloc(len);
let offset = 0;
@@ -182,9 +190,10 @@ function initSherpaOnnxOnlineModelConfig(config, Module) {
const modelTypeLen = Module.lengthBytesUTF8(config.modelType || '') + 1;
const modelingUnitLen = Module.lengthBytesUTF8(config.modelingUnit || '') + 1;
const bpeVocabLen = Module.lengthBytesUTF8(config.bpeVocab || '') + 1;
const tokensBufLen = Module.lengthBytesUTF8(config.tokensBuf || '') + 1;
const bufferLen =
tokensLen + providerLen + modelTypeLen + modelingUnitLen + bpeVocabLen;
const bufferLen = tokensLen + providerLen + modelTypeLen + modelingUnitLen +
bpeVocabLen + tokensBufLen;
const buffer = Module._malloc(bufferLen);
offset = 0;
@@ -204,6 +213,9 @@ function initSherpaOnnxOnlineModelConfig(config, Module) {
Module.stringToUTF8(config.bpeVocab || '', buffer + offset, bpeVocabLen);
offset += bpeVocabLen;
Module.stringToUTF8(config.tokensBuf || '', buffer + offset, tokensBufLen);
offset += tokensBufLen;
offset = transducer.len + paraformer.len + ctc.len;
Module.setValue(ptr + offset, buffer, 'i8*'); // tokens
offset += 4;
@@ -232,6 +244,16 @@ function initSherpaOnnxOnlineModelConfig(config, Module) {
'i8*'); // bpeVocab
offset += 4;
Module.setValue(
ptr + offset,
buffer + tokensLen + providerLen + modelTypeLen + modelingUnitLen +
bpeVocabLen,
'i8*'); // tokens_buf
offset += 4;
Module.setValue(ptr + offset, config.tokensBufSize || 0, 'i32');
offset += 4;
return {
buffer: buffer, ptr: ptr, len: len, transducer: transducer,
paraformer: paraformer, ctc: ctc
@@ -275,12 +297,20 @@ function initSherpaOnnxOnlineRecognizerConfig(config, Module) {
};
}
if (!('hotwordsBuf' in config)) {
config.hotwordsBuf = '';
}
if (!('hotwordsBufSize' in config)) {
config.hotwordsBufSize = 0;
}
const feat = initSherpaOnnxFeatureConfig(config.featConfig, Module);
const model = initSherpaOnnxOnlineModelConfig(config.modelConfig, Module);
const ctcFstDecoder = initSherpaOnnxOnlineCtcFstDecoderConfig(
config.ctcFstDecoderConfig, Module)
const len = feat.len + model.len + 8 * 4 + ctcFstDecoder.len + 3 * 4;
const len = feat.len + model.len + 8 * 4 + ctcFstDecoder.len + 5 * 4;
const ptr = Module._malloc(len);
let offset = 0;
@@ -295,8 +325,9 @@ function initSherpaOnnxOnlineRecognizerConfig(config, Module) {
const hotwordsFileLen = Module.lengthBytesUTF8(config.hotwordsFile || '') + 1;
const ruleFstsFileLen = Module.lengthBytesUTF8(config.ruleFsts || '') + 1;
const ruleFarsFileLen = Module.lengthBytesUTF8(config.ruleFars || '') + 1;
const bufferLen =
decodingMethodLen + hotwordsFileLen + ruleFstsFileLen + ruleFarsFileLen;
const hotwordsBufLen = Module.lengthBytesUTF8(config.hotwordsBuf || '') + 1;
const bufferLen = decodingMethodLen + hotwordsFileLen + ruleFstsFileLen +
ruleFarsFileLen + hotwordsBufLen;
const buffer = Module._malloc(bufferLen);
offset = 0;
@@ -314,6 +345,10 @@ function initSherpaOnnxOnlineRecognizerConfig(config, Module) {
Module.stringToUTF8(config.ruleFars || '', buffer + offset, ruleFarsFileLen);
offset += ruleFarsFileLen;
Module.stringToUTF8(
config.hotwordsBuf || '', buffer + offset, hotwordsBufLen);
offset += hotwordsBufLen;
offset = feat.len + model.len;
Module.setValue(ptr + offset, buffer, 'i8*'); // decoding method
offset += 4;
@@ -354,6 +389,16 @@ function initSherpaOnnxOnlineRecognizerConfig(config, Module) {
Module.setValue(ptr + offset, config.blankPenalty || 0, 'float');
offset += 4;
Module.setValue(
ptr + offset,
buffer + decodingMethodLen + hotwordsFileLen + ruleFstsFileLen +
ruleFarsFileLen,
'i8*');
offset += 4;
Module.setValue(ptr + offset, config.hotwordsBufSize || 0, 'i32');
offset += 4;
return {
buffer: buffer, ptr: ptr, len: len, feat: feat, model: model,
ctcFstDecoder: ctcFstDecoder

View File

@@ -19,14 +19,14 @@ static_assert(sizeof(SherpaOnnxOnlineZipformer2CtcModelConfig) == 1 * 4, "");
static_assert(sizeof(SherpaOnnxOnlineModelConfig) ==
sizeof(SherpaOnnxOnlineTransducerModelConfig) +
sizeof(SherpaOnnxOnlineParaformerModelConfig) +
sizeof(SherpaOnnxOnlineZipformer2CtcModelConfig) + 7 * 4,
sizeof(SherpaOnnxOnlineZipformer2CtcModelConfig) + 9 * 4,
"");
static_assert(sizeof(SherpaOnnxFeatureConfig) == 2 * 4, "");
static_assert(sizeof(SherpaOnnxOnlineCtcFstDecoderConfig) == 2 * 4, "");
static_assert(sizeof(SherpaOnnxOnlineRecognizerConfig) ==
sizeof(SherpaOnnxFeatureConfig) +
sizeof(SherpaOnnxOnlineModelConfig) + 8 * 4 +
sizeof(SherpaOnnxOnlineCtcFstDecoderConfig) + 3 * 4,
sizeof(SherpaOnnxOnlineCtcFstDecoderConfig) + 5 * 4,
"");
void MyPrint(SherpaOnnxOnlineRecognizerConfig *config) {
@@ -54,6 +54,9 @@ void MyPrint(SherpaOnnxOnlineRecognizerConfig *config) {
fprintf(stdout, "model type: %s\n", model_config->model_type);
fprintf(stdout, "modeling unit: %s\n", model_config->modeling_unit);
fprintf(stdout, "bpe vocab: %s\n", model_config->bpe_vocab);
fprintf(stdout, "tokens_buf: %s\n",
model_config->tokens_buf ? model_config->tokens_buf : "");
fprintf(stdout, "tokens_buf_size: %d\n", model_config->tokens_buf_size);
fprintf(stdout, "----------feat config----------\n");
fprintf(stdout, "sample rate: %d\n", feat->sample_rate);

View File

@@ -62,12 +62,20 @@ function initSherpaOnnxOnlineTransducerModelConfig(config, Module) {
// The user should free the returned pointers
function initModelConfig(config, Module) {
if (!('tokensBuf' in config)) {
config.tokensBuf = '';
}
if (!('tokensBufSize' in config)) {
config.tokensBufSize = 0;
}
const transducer =
initSherpaOnnxOnlineTransducerModelConfig(config.transducer, Module);
const paraformer_len = 2 * 4
const ctc_len = 1 * 4
const len = transducer.len + paraformer_len + ctc_len + 7 * 4;
const len = transducer.len + paraformer_len + ctc_len + 9 * 4;
const ptr = Module._malloc(len);
Module.HEAPU8.fill(0, ptr, ptr + len);
@@ -79,8 +87,9 @@ function initModelConfig(config, Module) {
const modelTypeLen = Module.lengthBytesUTF8(config.modelType || '') + 1;
const modelingUnitLen = Module.lengthBytesUTF8(config.modelingUnit || '') + 1;
const bpeVocabLen = Module.lengthBytesUTF8(config.bpeVocab || '') + 1;
const bufferLen =
tokensLen + providerLen + modelTypeLen + modelingUnitLen + bpeVocabLen;
const tokensBufLen = Module.lengthBytesUTF8(config.tokensBuf || '') + 1;
const bufferLen = tokensLen + providerLen + modelTypeLen + modelingUnitLen +
bpeVocabLen + tokensBufLen;
const buffer = Module._malloc(bufferLen);
offset = 0;
@@ -100,6 +109,9 @@ function initModelConfig(config, Module) {
Module.stringToUTF8(config.bpeVocab || '', buffer + offset, bpeVocabLen);
offset += bpeVocabLen;
Module.stringToUTF8(config.tokensBuf || '', buffer + offset, tokensBufLen);
offset += tokensBufLen;
offset = transducer.len + paraformer_len + ctc_len;
Module.setValue(ptr + offset, buffer, 'i8*'); // tokens
offset += 4;
@@ -128,6 +140,16 @@ function initModelConfig(config, Module) {
'i8*'); // bpeVocab
offset += 4;
Module.setValue(
ptr + offset,
buffer + tokensLen + providerLen + modelTypeLen + modelingUnitLen +
bpeVocabLen,
'i8*'); // tokens_buf
offset += 4;
Module.setValue(ptr + offset, config.tokensBufSize || 0, 'i32');
offset += 4;
return {
buffer: buffer, ptr: ptr, len: len, transducer: transducer
}

View File

@@ -19,7 +19,7 @@ static_assert(sizeof(SherpaOnnxOnlineZipformer2CtcModelConfig) == 1 * 4, "");
static_assert(sizeof(SherpaOnnxOnlineModelConfig) ==
sizeof(SherpaOnnxOnlineTransducerModelConfig) +
sizeof(SherpaOnnxOnlineParaformerModelConfig) +
sizeof(SherpaOnnxOnlineZipformer2CtcModelConfig) + 7 * 4,
sizeof(SherpaOnnxOnlineZipformer2CtcModelConfig) + 9 * 4,
"");
static_assert(sizeof(SherpaOnnxFeatureConfig) == 2 * 4, "");
static_assert(sizeof(SherpaOnnxKeywordSpotterConfig) ==