Pascal API for streaming ASR (#1246)
This commit is contained in:
Fangjun Kuang
9
pascal-api-examples/README.md
Normal file
9
pascal-api-examples/README.md
Normal file
@@ -0,0 +1,9 @@
|
||||
# Introduction
|
||||
|
||||
This directory contains examples for how to use the [Object Pascal](https://en.wikipedia.org/wiki/Object_Pascal)
|
||||
APIs of [sherpa-onnx](https://github.com/k2-fsa/sherpa-onnx).
|
||||
|
||||
|Directory| Description|
|
||||
|---------|------------|
|
||||
|[read-wav](./read-wav)|It shows how to read a wave file.|
|
||||
|[streaming-asr](./streaming-asr)| It shows how to use streaming models for speech recognition.|
|
||||
@@ -7,10 +7,11 @@ SHERPA_ONNX_DIR=$(cd $SCRIPT_DIR/../.. && pwd)
|
||||
|
||||
echo "SHERPA_ONNX_DIR: $SHERPA_ONNX_DIR"
|
||||
|
||||
if [[ ! -f ../../build/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/lib/libsherpa-onnx-c-api.so ]]; then
|
||||
if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
|
||||
mkdir -p ../../build
|
||||
pushd ../../build
|
||||
cmake \
|
||||
-DCMAKE_INSTALL_PREFIX=./install \
|
||||
-DSHERPA_ONNX_ENABLE_PYTHON=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_TESTS=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_CHECK=OFF \
|
||||
@@ -18,8 +19,7 @@ if [[ ! -f ../../build/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/lib/l
|
||||
-DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
|
||||
..
|
||||
|
||||
make -j4 sherpa-onnx-c-api
|
||||
ls -lh lib
|
||||
cmake --build . --target install --config Release
|
||||
popd
|
||||
fi
|
||||
|
||||
@@ -29,10 +29,10 @@ fi
|
||||
|
||||
fpc \
|
||||
-Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
|
||||
-Fl$SHERPA_ONNX_DIR/build/lib \
|
||||
-Fl$SHERPA_ONNX_DIR/build/install/lib \
|
||||
./main.pas
|
||||
|
||||
export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/lib:$LD_LIBRARY_PATH
|
||||
export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/lib:$DYLD_LIBRARY_PATH
|
||||
export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
|
||||
export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
|
||||
|
||||
./main
|
||||
|
||||
4
pascal-api-examples/streaming-asr/.gitignore
vendored
Normal file
4
pascal-api-examples/streaming-asr/.gitignore
vendored
Normal file
@@ -0,0 +1,4 @@
|
||||
zipformer_transducer
|
||||
paraformer
|
||||
zipformer_ctc
|
||||
zipformer_ctc_hlg
|
||||
11
pascal-api-examples/streaming-asr/README.md
Normal file
11
pascal-api-examples/streaming-asr/README.md
Normal file
@@ -0,0 +1,11 @@
|
||||
# Introduction
|
||||
|
||||
This folder contains examples about using sherpa-onnx's object pascal
|
||||
APIs with streaming models for speech recognition.
|
||||
|
||||
|File|Description|
|
||||
|----|-----------|
|
||||
|[run-paraformer.sh](./run-paraformer.sh)|Use a streaming Paraformer model for speech recognition|
|
||||
|[run-zipformer-ctc-hlg.sh](./run-zipformer-ctc-hlg.sh)|Use a streaming Zipformer CTC model for speech recognition|
|
||||
|[run-zipformer-ctc.sh](./run-zipformer-ctc.sh)|Use a streaming Zipformer CTC model with HLG for speech recognition|
|
||||
|[run-zipformer-transducer.sh](./run-zipformer-transducer.sh)|Use a Zipformer transducer model for speech recognition|
|
||||
88
pascal-api-examples/streaming-asr/paraformer.pas
Normal file
88
pascal-api-examples/streaming-asr/paraformer.pas
Normal file
@@ -0,0 +1,88 @@
|
||||
{ Copyright (c) 2024 Xiaomi Corporation }
|
||||
|
||||
{
|
||||
This file shows how to use a streaming Paraformer model to decode files.
|
||||
|
||||
You can download the model files from
|
||||
https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
}
|
||||
|
||||
program paraformer;
|
||||
|
||||
{$mode objfpc}
|
||||
|
||||
uses
|
||||
sherpa_onnx,
|
||||
DateUtils,
|
||||
SysUtils;
|
||||
|
||||
var
|
||||
Config: TSherpaOnnxOnlineRecognizerConfig;
|
||||
Recognizer: TSherpaOnnxOnlineRecognizer;
|
||||
Stream: TSherpaOnnxOnlineStream;
|
||||
RecognitionResult: TSherpaOnnxOnlineRecognizerResult;
|
||||
Wave: TSherpaOnnxWave;
|
||||
WaveFilename: AnsiString;
|
||||
TailPaddings: array of Single;
|
||||
|
||||
Start: TDateTime;
|
||||
Stop: TDateTime;
|
||||
|
||||
Elapsed: Single;
|
||||
Duration: Single;
|
||||
RealTimeFactor: Single;
|
||||
begin
|
||||
Initialize(Config);
|
||||
|
||||
{Please visit https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
to download model files used in this file.}
|
||||
Config.ModelConfig.Paraformer.Encoder := './sherpa-onnx-streaming-paraformer-bilingual-zh-en/encoder.int8.onnx';
|
||||
Config.ModelConfig.Paraformer.Decoder := './sherpa-onnx-streaming-paraformer-bilingual-zh-en/decoder.int8.onnx';
|
||||
Config.ModelConfig.Tokens := './sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt';
|
||||
|
||||
Config.ModelConfig.Provider := 'cpu';
|
||||
Config.ModelConfig.NumThreads := 1;
|
||||
Config.ModelConfig.Debug := False;
|
||||
|
||||
WaveFilename := './sherpa-onnx-streaming-paraformer-bilingual-zh-en/test_wavs/2.wav';
|
||||
|
||||
Wave := SherpaOnnxReadWave(WaveFilename);
|
||||
|
||||
Recognizer := TSherpaOnnxOnlineRecognizer.Create(Config);
|
||||
|
||||
Start := Now;
|
||||
|
||||
Stream := Recognizer.CreateStream();
|
||||
|
||||
Stream.AcceptWaveform(Wave.Samples, Wave.SampleRate);
|
||||
|
||||
SetLength(TailPaddings, Round(Wave.SampleRate * 0.5)); {0.5 seconds of padding}
|
||||
Stream.AcceptWaveform(TailPaddings, Wave.SampleRate);
|
||||
|
||||
Stream.InputFinished();
|
||||
|
||||
while Recognizer.IsReady(Stream) do
|
||||
Recognizer.Decode(Stream);
|
||||
|
||||
RecognitionResult := Recognizer.GetResult(Stream);
|
||||
|
||||
Stop := Now;
|
||||
|
||||
Elapsed := MilliSecondsBetween(Stop, Start) / 1000;
|
||||
Duration := Length(Wave.Samples) / Wave.SampleRate;
|
||||
RealTimeFactor := Elapsed / Duration;
|
||||
|
||||
WriteLn(RecognitionResult.ToString);
|
||||
WriteLn(Format('NumThreads %d', [Config.ModelConfig.NumThreads]));
|
||||
WriteLn(Format('Elapsed %.3f s', [Elapsed]));
|
||||
WriteLn(Format('Wave duration %.3f s', [Duration]));
|
||||
WriteLn(Format('RTF = %.3f/%.3f = %.3f', [Elapsed, Duration, RealTimeFactor]));
|
||||
|
||||
{Free resources to avoid memory leak.
|
||||
|
||||
Note: You don't need to invoke them for this simple script.
|
||||
However, you have to invoke them in your own large/complex project.
|
||||
}
|
||||
FreeAndNil(Stream);
|
||||
FreeAndNil(Recognizer);
|
||||
end.
|
||||
42
pascal-api-examples/streaming-asr/run-paraformer.sh
Executable file
42
pascal-api-examples/streaming-asr/run-paraformer.sh
Executable file
@@ -0,0 +1,42 @@
|
||||
#!/usr/bin/env bash
|
||||
|
||||
set -ex
|
||||
|
||||
SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
|
||||
SHERPA_ONNX_DIR=$(cd $SCRIPT_DIR/../.. && pwd)
|
||||
|
||||
echo "SHERPA_ONNX_DIR: $SHERPA_ONNX_DIR"
|
||||
|
||||
if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
|
||||
mkdir -p ../../build
|
||||
pushd ../../build
|
||||
cmake \
|
||||
-DCMAKE_INSTALL_PREFIX=./install \
|
||||
-DSHERPA_ONNX_ENABLE_PYTHON=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_TESTS=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_CHECK=OFF \
|
||||
-DBUILD_SHARED_LIBS=ON \
|
||||
-DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
|
||||
..
|
||||
|
||||
cmake --build . --target install --config Release
|
||||
ls -lh lib
|
||||
popd
|
||||
fi
|
||||
|
||||
|
||||
if [ ! -f ./sherpa-onnx-streaming-paraformer-bilingual-zh-en/tokens.txt ]; then
|
||||
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2
|
||||
tar xvf sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2
|
||||
rm sherpa-onnx-streaming-paraformer-bilingual-zh-en.tar.bz2
|
||||
fi
|
||||
|
||||
fpc \
|
||||
-Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
|
||||
-Fl$SHERPA_ONNX_DIR/build/install/lib \
|
||||
./paraformer.pas
|
||||
|
||||
export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
|
||||
export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
|
||||
|
||||
./paraformer
|
||||
41
pascal-api-examples/streaming-asr/run-zipformer-ctc-hlg.sh
Executable file
41
pascal-api-examples/streaming-asr/run-zipformer-ctc-hlg.sh
Executable file
@@ -0,0 +1,41 @@
|
||||
#!/usr/bin/env bash
|
||||
|
||||
set -ex
|
||||
|
||||
SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
|
||||
SHERPA_ONNX_DIR=$(cd $SCRIPT_DIR/../.. && pwd)
|
||||
|
||||
echo "SHERPA_ONNX_DIR: $SHERPA_ONNX_DIR"
|
||||
|
||||
if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
|
||||
mkdir -p ../../build
|
||||
pushd ../../build
|
||||
cmake \
|
||||
-DCMAKE_INSTALL_PREFIX=./install \
|
||||
-DSHERPA_ONNX_ENABLE_PYTHON=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_TESTS=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_CHECK=OFF \
|
||||
-DBUILD_SHARED_LIBS=ON \
|
||||
-DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
|
||||
..
|
||||
|
||||
cmake --build . --target install --config Release
|
||||
ls -lh lib
|
||||
popd
|
||||
fi
|
||||
|
||||
if [ ! -f ./sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/tokens.txt ]; then
|
||||
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
|
||||
tar xvf sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
|
||||
rm sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
|
||||
fi
|
||||
|
||||
fpc \
|
||||
-Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
|
||||
-Fl$SHERPA_ONNX_DIR/build/install/lib \
|
||||
./zipformer_ctc_hlg.pas
|
||||
|
||||
export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
|
||||
export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
|
||||
|
||||
./zipformer_ctc_hlg
|
||||
41
pascal-api-examples/streaming-asr/run-zipformer-ctc.sh
Executable file
41
pascal-api-examples/streaming-asr/run-zipformer-ctc.sh
Executable file
@@ -0,0 +1,41 @@
|
||||
#!/usr/bin/env bash
|
||||
|
||||
set -ex
|
||||
|
||||
SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
|
||||
SHERPA_ONNX_DIR=$(cd $SCRIPT_DIR/../.. && pwd)
|
||||
|
||||
echo "SHERPA_ONNX_DIR: $SHERPA_ONNX_DIR"
|
||||
|
||||
if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
|
||||
mkdir -p ../../build
|
||||
pushd ../../build
|
||||
cmake \
|
||||
-DCMAKE_INSTALL_PREFIX=./install \
|
||||
-DSHERPA_ONNX_ENABLE_PYTHON=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_TESTS=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_CHECK=OFF \
|
||||
-DBUILD_SHARED_LIBS=ON \
|
||||
-DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
|
||||
..
|
||||
|
||||
cmake --build . --target install --config Release
|
||||
ls -lh lib
|
||||
popd
|
||||
fi
|
||||
|
||||
if [ ! -f ./sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/tokens.txt ]; then
|
||||
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
|
||||
tar xvf sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
|
||||
rm sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18.tar.bz2
|
||||
fi
|
||||
|
||||
fpc \
|
||||
-Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
|
||||
-Fl$SHERPA_ONNX_DIR/build/install/lib \
|
||||
./zipformer_ctc.pas
|
||||
|
||||
export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
|
||||
export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
|
||||
|
||||
./zipformer_ctc
|
||||
42
pascal-api-examples/streaming-asr/run-zipformer-transducer.sh
Executable file
42
pascal-api-examples/streaming-asr/run-zipformer-transducer.sh
Executable file
@@ -0,0 +1,42 @@
|
||||
#!/usr/bin/env bash
|
||||
|
||||
set -ex
|
||||
|
||||
SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
|
||||
SHERPA_ONNX_DIR=$(cd $SCRIPT_DIR/../.. && pwd)
|
||||
|
||||
echo "SHERPA_ONNX_DIR: $SHERPA_ONNX_DIR"
|
||||
|
||||
if [[ ! -f ../../build/install/lib/libsherpa-onnx-c-api.dylib && ! -f ../../build/install/lib/libsherpa-onnx-c-api.so && ! -f ../../build/install/lib/sherpa-onnx-c-api.dll ]]; then
|
||||
mkdir -p ../../build
|
||||
pushd ../../build
|
||||
cmake \
|
||||
-DCMAKE_INSTALL_PREFIX=./install \
|
||||
-DSHERPA_ONNX_ENABLE_PYTHON=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_TESTS=OFF \
|
||||
-DSHERPA_ONNX_ENABLE_CHECK=OFF \
|
||||
-DBUILD_SHARED_LIBS=ON \
|
||||
-DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
|
||||
..
|
||||
|
||||
cmake --build . --target install --config Release
|
||||
ls -lh lib
|
||||
popd
|
||||
fi
|
||||
|
||||
if [ ! -f ./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/tokens.txt ]; then
|
||||
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20.tar.bz2
|
||||
tar xvf sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20.tar.bz2
|
||||
rm sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20.tar.bz2
|
||||
fi
|
||||
|
||||
|
||||
fpc \
|
||||
-Fu$SHERPA_ONNX_DIR/sherpa-onnx/pascal-api \
|
||||
-Fl$SHERPA_ONNX_DIR/build/install/lib \
|
||||
./zipformer_transducer.pas
|
||||
|
||||
export LD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$LD_LIBRARY_PATH
|
||||
export DYLD_LIBRARY_PATH=$SHERPA_ONNX_DIR/build/install/lib:$DYLD_LIBRARY_PATH
|
||||
|
||||
./zipformer_transducer
|
||||
87
pascal-api-examples/streaming-asr/zipformer_ctc.pas
Normal file
87
pascal-api-examples/streaming-asr/zipformer_ctc.pas
Normal file
@@ -0,0 +1,87 @@
|
||||
{ Copyright (c) 2024 Xiaomi Corporation }
|
||||
|
||||
{
|
||||
This file shows how to use a streaming Zipformer CTC model
|
||||
to decode files.
|
||||
|
||||
You can download the model files from
|
||||
https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
}
|
||||
|
||||
program zipformer_ctc;
|
||||
|
||||
{$mode objfpc}
|
||||
|
||||
uses
|
||||
sherpa_onnx,
|
||||
DateUtils,
|
||||
SysUtils;
|
||||
|
||||
var
|
||||
Config: TSherpaOnnxOnlineRecognizerConfig;
|
||||
Recognizer: TSherpaOnnxOnlineRecognizer;
|
||||
Stream: TSherpaOnnxOnlineStream;
|
||||
RecognitionResult: TSherpaOnnxOnlineRecognizerResult;
|
||||
Wave: TSherpaOnnxWave;
|
||||
WaveFilename: AnsiString;
|
||||
TailPaddings: array of Single;
|
||||
|
||||
Start: TDateTime;
|
||||
Stop: TDateTime;
|
||||
|
||||
Elapsed: Single;
|
||||
Duration: Single;
|
||||
RealTimeFactor: Single;
|
||||
begin
|
||||
Initialize(Config);
|
||||
|
||||
{Please visit https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
to download model files used in this file.}
|
||||
Config.ModelConfig.Zipformer2Ctc.Model := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/ctc-epoch-30-avg-3-chunk-16-left-128.int8.onnx';
|
||||
Config.ModelConfig.Tokens := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/tokens.txt';
|
||||
Config.ModelConfig.Provider := 'cpu';
|
||||
Config.ModelConfig.NumThreads := 1;
|
||||
Config.ModelConfig.Debug := False;
|
||||
|
||||
WaveFilename := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/test_wavs/8k.wav';
|
||||
|
||||
Wave := SherpaOnnxReadWave(WaveFilename);
|
||||
|
||||
Recognizer := TSherpaOnnxOnlineRecognizer.Create(Config);
|
||||
|
||||
Start := Now;
|
||||
|
||||
Stream := Recognizer.CreateStream();
|
||||
|
||||
Stream.AcceptWaveform(Wave.Samples, Wave.SampleRate);
|
||||
|
||||
SetLength(TailPaddings, Round(Wave.SampleRate * 0.5)); {0.5 seconds of padding}
|
||||
Stream.AcceptWaveform(TailPaddings, Wave.SampleRate);
|
||||
|
||||
Stream.InputFinished();
|
||||
|
||||
while Recognizer.IsReady(Stream) do
|
||||
Recognizer.Decode(Stream);
|
||||
|
||||
RecognitionResult := Recognizer.GetResult(Stream);
|
||||
|
||||
Stop := Now;
|
||||
|
||||
Elapsed := MilliSecondsBetween(Stop, Start) / 1000;
|
||||
Duration := Length(Wave.Samples) / Wave.SampleRate;
|
||||
RealTimeFactor := Elapsed / Duration;
|
||||
|
||||
WriteLn(RecognitionResult.ToString);
|
||||
WriteLn(Format('NumThreads %d', [Config.ModelConfig.NumThreads]));
|
||||
WriteLn(Format('Elapsed %.3f s', [Elapsed]));
|
||||
WriteLn(Format('Wave duration %.3f s', [Duration]));
|
||||
WriteLn(Format('RTF = %.3f/%.3f = %.3f', [Elapsed, Duration, RealTimeFactor]));
|
||||
|
||||
{Free resources to avoid memory leak.
|
||||
|
||||
Note: You don't need to invoke them for this simple script.
|
||||
However, you have to invoke them in your own large/complex project.
|
||||
}
|
||||
FreeAndNil(Stream);
|
||||
FreeAndNil(Recognizer);
|
||||
end.
|
||||
88
pascal-api-examples/streaming-asr/zipformer_ctc_hlg.pas
Normal file
88
pascal-api-examples/streaming-asr/zipformer_ctc_hlg.pas
Normal file
@@ -0,0 +1,88 @@
|
||||
{ Copyright (c) 2024 Xiaomi Corporation }
|
||||
|
||||
{
|
||||
This file shows how to use a streaming Zipformer CTC model
|
||||
with HLG to decode files.
|
||||
|
||||
You can download the model files from
|
||||
https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
}
|
||||
|
||||
program zipformer_ctc_hlg;
|
||||
|
||||
{$mode objfpc}
|
||||
|
||||
uses
|
||||
sherpa_onnx,
|
||||
DateUtils,
|
||||
SysUtils;
|
||||
|
||||
var
|
||||
Config: TSherpaOnnxOnlineRecognizerConfig;
|
||||
Recognizer: TSherpaOnnxOnlineRecognizer;
|
||||
Stream: TSherpaOnnxOnlineStream;
|
||||
RecognitionResult: TSherpaOnnxOnlineRecognizerResult;
|
||||
Wave: TSherpaOnnxWave;
|
||||
WaveFilename: AnsiString;
|
||||
TailPaddings: array of Single;
|
||||
|
||||
Start: TDateTime;
|
||||
Stop: TDateTime;
|
||||
|
||||
Elapsed: Single;
|
||||
Duration: Single;
|
||||
RealTimeFactor: Single;
|
||||
begin
|
||||
Initialize(Config);
|
||||
|
||||
{Please visit https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
to download model files used in this file.}
|
||||
Config.ModelConfig.Zipformer2Ctc.Model := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/ctc-epoch-30-avg-3-chunk-16-left-128.int8.onnx';
|
||||
Config.ModelConfig.Tokens := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/tokens.txt';
|
||||
Config.ModelConfig.Provider := 'cpu';
|
||||
Config.ModelConfig.NumThreads := 1;
|
||||
Config.ModelConfig.Debug := True;
|
||||
Config.CtcFstDecoderConfig.Graph := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/HLG.fst';
|
||||
|
||||
WaveFilename := './sherpa-onnx-streaming-zipformer-ctc-small-2024-03-18/test_wavs/8k.wav';
|
||||
|
||||
Wave := SherpaOnnxReadWave(WaveFilename);
|
||||
|
||||
Recognizer := TSherpaOnnxOnlineRecognizer.Create(Config);
|
||||
|
||||
Start := Now;
|
||||
|
||||
Stream := Recognizer.CreateStream();
|
||||
|
||||
Stream.AcceptWaveform(Wave.Samples, Wave.SampleRate);
|
||||
|
||||
SetLength(TailPaddings, Round(Wave.SampleRate * 0.5)); {0.5 seconds of padding}
|
||||
Stream.AcceptWaveform(TailPaddings, Wave.SampleRate);
|
||||
|
||||
Stream.InputFinished();
|
||||
|
||||
while Recognizer.IsReady(Stream) do
|
||||
Recognizer.Decode(Stream);
|
||||
|
||||
RecognitionResult := Recognizer.GetResult(Stream);
|
||||
|
||||
Stop := Now;
|
||||
|
||||
Elapsed := MilliSecondsBetween(Stop, Start) / 1000;
|
||||
Duration := Length(Wave.Samples) / Wave.SampleRate;
|
||||
RealTimeFactor := Elapsed / Duration;
|
||||
|
||||
WriteLn(RecognitionResult.ToString);
|
||||
WriteLn(Format('NumThreads %d', [Config.ModelConfig.NumThreads]));
|
||||
WriteLn(Format('Elapsed %.3f s', [Elapsed]));
|
||||
WriteLn(Format('Wave duration %.3f s', [Duration]));
|
||||
WriteLn(Format('RTF = %.3f/%.3f = %.3f', [Elapsed, Duration, RealTimeFactor]));
|
||||
|
||||
{Free resources to avoid memory leak.
|
||||
|
||||
Note: You don't need to invoke them for this simple script.
|
||||
However, you have to invoke them in your own large/complex project.
|
||||
}
|
||||
FreeAndNil(Stream);
|
||||
FreeAndNil(Recognizer);
|
||||
end.
|
||||
89
pascal-api-examples/streaming-asr/zipformer_transducer.pas
Normal file
89
pascal-api-examples/streaming-asr/zipformer_transducer.pas
Normal file
@@ -0,0 +1,89 @@
|
||||
{ Copyright (c) 2024 Xiaomi Corporation }
|
||||
|
||||
{
|
||||
This file shows how to use a streaming Zipformer transducer
|
||||
to decode files.
|
||||
|
||||
You can download the model files from
|
||||
https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
}
|
||||
|
||||
program zipformer_transducer;
|
||||
|
||||
{$mode objfpc}
|
||||
|
||||
uses
|
||||
sherpa_onnx,
|
||||
DateUtils,
|
||||
SysUtils;
|
||||
|
||||
var
|
||||
Config: TSherpaOnnxOnlineRecognizerConfig;
|
||||
Recognizer: TSherpaOnnxOnlineRecognizer;
|
||||
Stream: TSherpaOnnxOnlineStream;
|
||||
RecognitionResult: TSherpaOnnxOnlineRecognizerResult;
|
||||
Wave: TSherpaOnnxWave;
|
||||
WaveFilename: AnsiString;
|
||||
TailPaddings: array of Single;
|
||||
|
||||
Start: TDateTime;
|
||||
Stop: TDateTime;
|
||||
|
||||
Elapsed: Single;
|
||||
Duration: Single;
|
||||
RealTimeFactor: Single;
|
||||
begin
|
||||
Initialize(Config);
|
||||
|
||||
{Please visit https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
|
||||
to download model files used in this file.}
|
||||
Config.ModelConfig.Transducer.Encoder := './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/encoder-epoch-99-avg-1.int8.onnx';
|
||||
Config.ModelConfig.Transducer.Decoder := './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/decoder-epoch-99-avg-1.onnx';
|
||||
Config.ModelConfig.Transducer.Joiner := './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/joiner-epoch-99-avg-1.int8.onnx';
|
||||
Config.ModelConfig.Tokens := './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/tokens.txt';
|
||||
Config.ModelConfig.Provider := 'cpu';
|
||||
Config.ModelConfig.NumThreads := 1;
|
||||
Config.ModelConfig.Debug := False;
|
||||
|
||||
WaveFilename := './sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/test_wavs/0.wav';
|
||||
|
||||
Wave := SherpaOnnxReadWave(WaveFilename);
|
||||
|
||||
Recognizer := TSherpaOnnxOnlineRecognizer.Create(Config);
|
||||
|
||||
Start := Now;
|
||||
|
||||
Stream := Recognizer.CreateStream();
|
||||
|
||||
Stream.AcceptWaveform(Wave.Samples, Wave.SampleRate);
|
||||
|
||||
SetLength(TailPaddings, Round(Wave.SampleRate * 0.5)); {0.5 seconds of padding}
|
||||
Stream.AcceptWaveform(TailPaddings, Wave.SampleRate);
|
||||
|
||||
Stream.InputFinished();
|
||||
|
||||
while Recognizer.IsReady(Stream) do
|
||||
Recognizer.Decode(Stream);
|
||||
|
||||
RecognitionResult := Recognizer.GetResult(Stream);
|
||||
|
||||
Stop := Now;
|
||||
|
||||
Elapsed := MilliSecondsBetween(Stop, Start) / 1000;
|
||||
Duration := Length(Wave.Samples) / Wave.SampleRate;
|
||||
RealTimeFactor := Elapsed / Duration;
|
||||
|
||||
WriteLn(RecognitionResult.ToString);
|
||||
WriteLn(Format('NumThreads %d', [Config.ModelConfig.NumThreads]));
|
||||
WriteLn(Format('Elapsed %.3f s', [Elapsed]));
|
||||
WriteLn(Format('Wave duration %.3f s', [Duration]));
|
||||
WriteLn(Format('RTF = %.3f/%.3f = %.3f', [Elapsed, Duration, RealTimeFactor]));
|
||||
|
||||
{Free resources to avoid memory leak.
|
||||
|
||||
Note: You don't need to invoke them for this simple script.
|
||||
However, you have to invoke them in your own large/complex project.
|
||||
}
|
||||
FreeAndNil(Stream);
|
||||
FreeAndNil(Recognizer);
|
||||
end.
|
||||
Reference in New Issue
Block a user