Skip to content
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

Removed continuous recognition example and replaced it with a link the echosharp #309

Merged
merged 1 commit into from
Dec 26, 2024
Merged

Removed continuous recognition example and replaced it with a link the echosharp #309

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
143 changes: 2 additions & 141 deletions examples/ContinuousRecognition/Program.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,143 +1,4 @@
// Licensed under the MIT license: https://opensource.org/licenses/MIT

using Whisper.net;
using Whisper.net.Ggml;
using Whisper.net.Wave;

var ggmlType = GgmlType.TinyEn;
var modelFileName = "ggml-tinyen.bin";
var wavFileName = "bush.wav";

var maxProcessingTimeMs = 10000;
var minProcessingTimeMs = 1500;
var advancingProcessingTimeMs = 500;

if (!File.Exists(modelFileName))
{
await DownloadModel(modelFileName, ggmlType);
}

using var whisperFactory = WhisperFactory.FromPath(modelFileName);

var builder = whisperFactory.CreateBuilder()
.WithProbabilities()
.WithLanguage("en");

using var fileStream = File.OpenRead(wavFileName);
var waveParser = new WaveParser(fileStream);
await waveParser.InitializeAsync();

var samples = new float[waveParser.SampleRate / 1000 * maxProcessingTimeMs];

// Process first the minimum processing time of the audio file

// Read first min processing time into samples
var dataPosition = waveParser.DataChunkPosition;

fileStream.Seek(dataPosition, SeekOrigin.Begin);

var partialResults = new List<(List<SegmentData> segments, TimeSpan startTime, TimeSpan endTime)>();
var buffer = new byte[waveParser.SampleRate / 1000 * maxProcessingTimeMs * 2 * waveParser.Channels];

var bufferSize = waveParser.SampleRate / 1000 * minProcessingTimeMs * 2 * waveParser.Channels;

var bytesRead = await fileStream.ReadAsync(buffer.AsMemory(0, (int)bufferSize));

var currentSampleIndex = 0;

for (var i = 0; i < bytesRead;)
{
long sampleSum = 0;

for (var currentChannel = 0; currentChannel < waveParser.Channels; currentChannel++)
{
sampleSum += BitConverter.ToInt16(buffer, i);
i += 2;
}

samples[currentSampleIndex++] = sampleSum / (float)waveParser.Channels / 32768.0f;
}

var currentProcessedStartTime = TimeSpan.Zero;
var currentProcessedEndTime = TimeSpan.FromMilliseconds(minProcessingTimeMs);

await using (var processor = builder.Build())
{
var segments = new List<SegmentData>();
await foreach (var data in processor.ProcessAsync(samples.AsMemory(0, currentSampleIndex)))
{
segments.Add(data);

}
partialResults.Add((segments, currentProcessedStartTime, currentProcessedEndTime));
}

var fullText = string.Empty;

while (currentSampleIndex < waveParser.SamplesCount)
{
bufferSize = waveParser.SampleRate / 1000 * advancingProcessingTimeMs * 2 * waveParser.Channels;

bytesRead = await fileStream.ReadAsync(buffer.AsMemory(0, (int)bufferSize));
for (var i = 0; i < bytesRead;)
{
long sampleSum = 0;

for (var currentChannel = 0; currentChannel < waveParser.Channels; currentChannel++)
{
sampleSum += BitConverter.ToInt16(buffer, i);
i += 2;
}

samples[currentSampleIndex++] = sampleSum / (float)waveParser.Channels / 32768.0f;
}

currentProcessedEndTime = currentProcessedEndTime.Add(TimeSpan.FromMilliseconds(advancingProcessingTimeMs));

await using (var processor = builder.Build())
{
var segments = new List<SegmentData>();
await foreach (var data in processor.ProcessAsync(samples.AsMemory(0, currentSampleIndex)))
{
segments.Add(data);
}
partialResults.Add((segments, currentProcessedStartTime, currentProcessedEndTime));

var indexSegment = 0;
foreach (var segment in segments)
{
Console.WriteLine($"{indexSegment}: {segment.Start}->{segment.End}: {segment.Text} => with probability: {segment.Probability}");
indexSegment++;
}
}

var indexPartial = 0;
//TODO: Check if partials concluded to one finished segment and return it.
foreach (var partial in partialResults)
{
// Console.WriteLine(indexPartial + ":" + partial.startTime + " - " + partial.endTime + " " + partial.segments.Count + " segments\n-----------");
indexPartial++;
// If one segment is identified. E.g. "My fellow Americans" from second 0 to second 3 => we remove that part from the samples, adding the text to the prompt and continue processing the rest of the samples.
}

// If the total current processing time is reaching max processing time => we remove half of the samples and continue processing the rest of the samples.
if (currentProcessedEndTime.TotalMilliseconds - currentProcessedStartTime.TotalMilliseconds >= maxProcessingTimeMs)
{
// First, we copy the last part of the samples to the beginning of the array
var samplesToCopy = currentSampleIndex - maxProcessingTimeMs / 2;
for (var i = 0; i < samplesToCopy; i++)
{
samples[i] = samples[i + maxProcessingTimeMs / 2];
}
currentProcessedStartTime = currentProcessedStartTime.Add(TimeSpan.FromMilliseconds(maxProcessingTimeMs / 2));
currentSampleIndex = samplesToCopy;
}
}

static async Task DownloadModel(string fileName, GgmlType ggmlType)
{
Console.WriteLine($"Downloading Model {fileName}");
using var modelStream = await WhisperGgmlDownloader.GetGgmlModelAsync(ggmlType);
using var fileWriter = File.OpenWrite(fileName);
await modelStream.CopyToAsync(fileWriter);
}
// This example was moved to echo sharp to allow better real-time transcription using VAD components, see: https://github.com/sandrohanea/echosharp
Console.WriteLine("Moved to https://github.com/sandrohanea/echosharp");
Loading