public void TestProcess()
{
using Cheetah cheetah = Cheetah.Create(ACCESS_KEY, null, 0.2f);
string testAudioPath = Path.Combine(_relativeDir, "resources/audio_samples/test.wav");
List <short> pcm = GetPcmFromFile(testAudioPath, cheetah.SampleRate);
int frameLen = cheetah.FrameLength;
int framecount = (int)Math.Floor((float)(pcm.Count / frameLen));
string transcript = "";
bool isEndpoint = false;
for (int i = 0; i < framecount; i++)
{
int start = i * cheetah.FrameLength;
List <short> frame = pcm.GetRange(start, frameLen);
CheetahTranscript transcriptObj = cheetah.Process(frame.ToArray());
transcript += transcriptObj.Transcript;
isEndpoint = transcriptObj.IsEndpoint;
}
CheetahTranscript finalTranscriptObj = cheetah.Flush();
transcript += finalTranscriptObj.Transcript;
Assert.AreEqual(transcript, REF_TRANSCRIPT);
Assert.IsTrue(isEndpoint);
}
public void TestCustomModel()
{
string testModelPath = Path.Combine(_relativeDir, "lib/common/cheetah_params.pv");
using Cheetah cheetah = Cheetah.Create(ACCESS_KEY, testModelPath);
string testAudioPath = Path.Combine(_relativeDir, "resources/audio_samples/test.wav");
List <short> pcm = GetPcmFromFile(testAudioPath, cheetah.SampleRate);
int frameLen = cheetah.FrameLength;
int framecount = (int)Math.Floor((float)(pcm.Count / frameLen));
string transcript = "";
for (int i = 0; i < framecount; i++)
{
int start = i * cheetah.FrameLength;
List <short> frame = pcm.GetRange(start, frameLen);
CheetahTranscript transcriptObj = cheetah.Process(frame.ToArray());
transcript += transcriptObj.Transcript;
}
CheetahTranscript finalTranscriptObj = cheetah.Flush();
transcript += finalTranscriptObj.Transcript;
Assert.AreEqual(transcript, REF_TRANSCRIPT);
}
/// <summary>
/// Creates an input audio stream and instantiates an instance of Cheetah object.
/// </summary>
/// <param name="accessKey">AccessKey obtained from Picovoice Console (https://console.picovoice.ai/).</param>
/// <param name="modelPath">Absolute path to the file containing model parameters. If not set it will be set to the default location.</param>
/// <param name="endpointDurationSec">
/// Duration of endpoint in seconds. A speech endpoint is detected when there is a segment of audio(with a duration specified herein) after
/// an utterance without any speech in it. Set to `0` to disable
/// </param>
/// <param name="audioDeviceIndex">Optional argument. If provided, audio is recorded from this input device. Otherwise, the default audio input device is used.</param>
public static void RunDemo(
string accessKey,
string modelPath,
float endpointDurationSec,
int audioDeviceIndex)
{
Cheetah cheetah = null;
cheetah = Cheetah.Create(
accessKey,
modelPath,
endpointDurationSec);
PvRecorder recorder = PvRecorder.Create(audioDeviceIndex, cheetah.FrameLength);
recorder.Start();
List <short> audioFrame = new List <short>();
Console.CancelKeyPress += (s, o) =>
{
Console.WriteLine("Stopping...");
};
Console.WriteLine($"\nUsing device: {recorder.SelectedDevice}");
Console.WriteLine(">>> Press `CTRL+C` to exit:\n");
try
{
while (true)
{
short[] pcm = recorder.Read();
CheetahTranscript transcriptObj = cheetah.Process(pcm);
if (!String.IsNullOrEmpty(transcriptObj.Transcript))
{
Console.Write(transcriptObj.Transcript);
}
if (transcriptObj.IsEndpoint)
{
CheetahTranscript finalTranscriptObj = cheetah.Flush();
Console.WriteLine(finalTranscriptObj.Transcript);
}
}
}
catch (CheetahActivationLimitException)
{
Console.WriteLine($"AccessKey '{accessKey}' has reached its processing limit.");
}
finally
{
cheetah.Dispose();
recorder.Dispose();
}
}
/// <summary>
/// Reads through input file and prints the transcription returned by Cheetah.
/// </summary>
/// <param name="inputAudioPath">Required argument. Absolute path to input audio file.</param>
/// <param name="accessKey">AccessKey obtained from Picovoice Console (https://console.picovoice.ai/).</param>
/// <param name="modelPath">Absolute path to the file containing model parameters. If not set it will be set to the default location.</param>
/// <param name="endpointDurationSec">
/// Duration of endpoint in seconds. A speech endpoint is detected when there is a segment of audio(with a duration specified herein) after
/// an utterance without any speech in it. Set to `0` to disable
/// </param>
/// </param>
public static void RunDemo(
string accessKey,
string inputAudioPath,
string modelPath)
{
// init Cheetah speech-to-text engine
using Cheetah cheetah = Cheetah.Create(
accessKey,
modelPath);
using BinaryReader reader = new BinaryReader(File.Open(inputAudioPath, FileMode.Open));
ValidateWavFile(reader, cheetah.SampleRate, 16, out short numChannels);
short[] cheetahFrame = new short[cheetah.FrameLength];
int frameIndex = 0;
while (reader.BaseStream.Position != reader.BaseStream.Length)
{
cheetahFrame[frameIndex++] = reader.ReadInt16();
if (frameIndex == cheetahFrame.Length)
{
try
{
CheetahTranscript transcriptObj = cheetah.Process(cheetahFrame);
if (!String.IsNullOrEmpty(transcriptObj.Transcript))
{
Console.Write(transcriptObj.Transcript);
}
}
catch (CheetahActivationLimitException)
{
cheetah.Dispose();
Console.WriteLine($"AccessKey '{accessKey}' has reached its processing limit.");
}
frameIndex = 0;
}
// skip right channel
if (numChannels == 2)
{
reader.ReadInt16();
}
}
CheetahTranscript finalTranscriptObj = cheetah.Flush();
string transcript = finalTranscriptObj.Transcript;
if (!String.IsNullOrEmpty(transcript))
{
Console.WriteLine(transcript);
}
}