Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.
Python asraudiochineseemotion-recognitionfunasrmcp-servermultilingual-asropenai-compatible-apiparaformerpunctuationpytorchreal-time-asrspeaker-diarizationspeech-recognitionspeech-to-textstreaming-asrtranscriptionvllmvoice-activity-detectionwhisper-alternative
#include "result-json.h"
#include <cassert>
#include <stdexcept>
#include <string>
#include "nlohmann/json.hpp"
int main() {
using funasr::FormatResultJson;
using funasr::OutputFormat;
using funasr::ParseOutputFormat;
assert(ParseOutputFormat("log") == OutputFormat::kLog);
assert(ParseOutputFormat("jsonl") == OutputFormat::kJsonl);
bool rejected_invalid_format = false;
try {
ParseOutputFormat("yaml");
} catch (const std::invalid_argument&) {
rejected_invalid_format = true;
}
assert(rejected_invalid_format);
const auto record = nlohmann::json::parse(FormatResultJson(
"utt-\"1", "offline", "你好 \"FunASR\"", "[[0,320],[320,660]]",
"[{\"text_seg\":\"你好\",\"start\":0,\"end\":660}]"));
assert(record.at("key") == "utt-\"1");
assert(record.at("mode") == "offline");
assert(record.at("text") == "你好 \"FunASR\"");
assert(record.at("timestamp") == nlohmann::json::parse("[[0,320],[320,660]]"));
assert(record.at("stamp_sents").at(0).at("start") == 0);
const auto empty_record = nlohmann::json::parse(
FormatResultJson("utt-2", "2pass", "done", "", "not-json"));
assert(empty_record.at("timestamp") == nlohmann::json::array());
assert(empty_record.at("stamp_sents") == nlohmann::json::array());
const auto wrong_type_record = nlohmann::json::parse(
FormatResultJson("utt-3", "online", "partial", "{\"start\":0}", "null"));
assert(wrong_type_record.at("timestamp") == nlohmann::json::array());
assert(wrong_type_record.at("stamp_sents") == nlohmann::json::array());
return 0;
}