picovoice / porcupine Goto Github PK

I have problem with how to import porcupine to my python script. I had many problems with it, that i said that it could not import the library and so on, but that is fixed.

My code:
import sys
import soundfile
import os
import pyaudio

sys.path.append(r'E:/Desktop/Python/WakeWord/Porcupine/binding/python/')

from porcupine import Porcupine

library_path = 'Porcupine/lib/windows/amd64/libpv_porcupine.dll'
model_file_path = 'Porcupine/lib/common/porcupine_params.pv'
keyword_file_paths = 'Porcupine/resources/keyword_files/alexa_windows.ppn'
sensitivities = [0.5]
handle = Porcupine(library_path, model_file_path, keyword_file_paths=keyword_file_paths, sensitivities=sensitivities)

def get_next_audio_frame():
pass

while True:
pcm = get_next_audio_frame()
keyword_index = handle.process(pcm)
if keyword_index >= 0:
# detection event logic/callback
pass

And my error:
Traceback (most recent call last): File "program.py", line 15, in <module> handle = Porcupine(library_path, model_file_path, keyword_file_paths=keyword_file_paths, sensitivities=sensitivities) File "E:/Desktop/Python/WakeWord/Porcupine/binding/python\porcupine.py", line 84, in __init__ raise ValueError("Different number of sensitivity and keyword file path parameters are provided.") ValueError: Different number of sensitivity and keyword file path parameters are provided.

HI,

[ERROR] could not find the pronunciation for 'trello'.

Is it possible to add this word to the dictionary?

I've been trying for the past 5 hours. Any tutorial on how to import it?

Expected behavior

The python demo keeps running until I quit it.

Actual behavior

The demo crashed with following error:

Traceback (most recent call last):
  File "demo/python/porcupine_demo.py", line 207, in <module>
    input_device_index=args.input_audio_device_index).run()
  File "demo/python/porcupine_demo.py", line 104, in run
    pcm = audio_stream.read(porcupine.frame_length)
  File "/usr/lib/python2.7/dist-packages/pyaudio.py", line 608, in read
    return pa.read_stream(self._stream, num_frames, exception_on_overflow)
IOError: [Errno -9981] Input overflowed

Steps to reproduce the behavior

Start the demo on a Rasperry PI Zero with ReSpeaker 2-Mic-HAT with following settings:

python demo/python/porcupine_demo.py --keyword_file_paths ./resources/keyword_files/grasshopper_raspberrypi.ppn --library_path ./lib/raspberry-pi/arm11/libpv_porcupine.so --input_audio_device_index 2

OS is Raspbian Stretch Lite.

Hey, I wanted to use this in a c++ program, I can include the headers but have no idea how to link the dll.
I've tried to look how to link it using CMake for a couple of days now with no success.
I get
C:/Users/tatan/CLionProjects/TreeVoiceAssistant/HotwordDetection.cpp:27: undefined reference to pv_porcupine_process(pv_porcupine_object*, short const*, bool*)'`
every single time.
I'm using CLion as you can probably see, with MINGW and I can sucessfuly include the headers, but dlls just won't.

Please help

Expected behavior

tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources/ -w "ok google" -p raspberrypi -o .

outputs a .ppn file

Actual behavior

[WARN] This version of optimizer cannot create keyword files for raspberrypi. Please contact [email protected].

Steps to reproduce the behavior

$ tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources/ -w "ok google" -p raspberrypi -o .

Please include the operating system and CPU architecture. When applicable, provide relevant audio data.

Expected behavior

Train optimizer with "homie"

Actual behavior

[ERROR] could not find the pronunciation for 'homie'

Steps to reproduce the behavior

tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources -w homie -p mac -o my_models

Please include enough details so that the issue can reproduced independently by the resolver.

Error

I tried to first make a wake word. By using your optimizer, and it worked. And then I imported it. But this time it says that a object is a NoneType... I don't know what that means, but I google it and find out it maybe a empty variable. I don't know the cuase of this problem, so hope you guys know it. My code is this:

import os
import sys

sys.path.append(os.path.join(os.path.dirname(__file__), 'Porcupine/binding/python/'))

from porcupine import Porcupine

library_path = "Porcupine/lib/windows/amd64/libpv_porcupine.dll"
model_file_path = "Porcupine/lib/common/porcupine_params.pv"
keyword_file_paths = ['Porcupine/jarvis_windows.ppn']
sensitivities = [0.5]
handle = Porcupine(library_path, model_file_path, keyword_file_paths=keyword_file_paths, sensitivities=sensitivities)

def get_next_audio_frame():
    pass

while True:
    pcm = get_next_audio_frame()
    keyword_index = handle.process(pcm)
    if keyword_index >= 0:
        # detection event logic/callback
        pass

handle.delete()`

And the error im getting:
`Traceback (most recent call last):
  File "program.py", line 19, in <module>
    keyword_index = handle.process(pcm)
  File "Porcupine/binding/python\porcupine.py", line 154, in process
    status = self.process_func(self._handle, (c_short * len(pcm))(*pcm), byref(result))
TypeError: object of type 'NoneType' has no len()

I also gave you guys a picture of my file dir. So you can see where the files is.

Hope you guys can fix it... Just contact me over this issue, or via my email :)

• Fredrik

so i created a ppn file for mac x86_64
"hey janet"

even with the sensitivities set to [0.1] it picks up odd words when i say things like
"hey jant"
"hey jan Bert" (and i really make sure the Bert sound comes out strong)

i even tried setting the sen to [0.00001] and it still fires with odd sounds

i would have thought the sensitivity closer to 1 would false positive lots.. but even 0.001 it false positives with close sounding words.. but "jan BERT" really has a B sound which i would have expected with a low sensitivity to not accept

Expected behavior

I expect to see a positive result when saying the wake word when running the demo on Windows.

Actual behavior

No result is reported.

Steps to reproduce the behavior

on a windows machine run:
python demo/python/porcupine_demo.py --keyword_file_paths alexa_windows.ppn

Please include the operating system and CPU architecture. When applicable, provide relevant audio data. Windows 10, Intel Core i7

Expected behavior

Install instructions needed.

Actual behavior

none

Steps to reproduce the behavior

I want to install it on a RPI but cannot see any instructions on how to do so
Please include the operating system and CPU architecture. When applicable, provide relevant audio data.
RPI 3 B+ running debian stretch

Expected behavior

I downloaded project .zip file from github, and tried to execute demo/andriod/app in Android Studio.
However, Gradle project sync failed.
Please help me!

Actual behavior

Error messages are below
Unable to resolve dependency for ':porcupinemanager@debug/compileClasspath': Could not resolve project :porcupine.
Unable to resolve dependency for ':porcupinemanager@debugAndroidTest/compileClasspath': Could not resolve project :porcupine.
Unable to resolve dependency for ':porcupinemanager@debugUnitTest/compileClasspath': Could not resolve project :porcupine.
Unable to resolve dependency for ':porcupinemanager@release/compileClasspath': Could not resolve project :porcupine.
Unable to resolve dependency for ':porcupinemanager@releaseUnitTest/compileClasspath': Could not resolve project :porcupine.

Steps to reproduce the behavior

download zip file and execute in Android Studio

Please include enough details so that the issue can reproduced independently by the resolver.

Hi Alireza,
We ran into a porcupine crash, when we tried creating multiple porcupine handles for processing multiple audio streams simultaneously.

Steps to reproduce:

1. The C/C++ Application spawns two threads and creates two porcupine handles(handle1 and handle2) for detecting keywords from two audio steams.

2. The first thread finishes detection for the first audio stream and frees the porcupine handle using the method pv_porcupine_delete(handle1);

3. But the second thread still continues feeding audio data to porcupine engine from second audio stream using handle2.
   Here after the 3rd step, the application crashes with the following back-trace…

      #######
      6505== Invalid read of size 4
      6505==    at 0x40666A8: pv_sqrt (in /root/Arun/Porcupine-master/lib/linux/i386/libpv_porcupine.so)
      6505==    by 0x406439C: pv_specgram_compute (in /root/Arun/Porcupine-master/lib/linux/i386/libpv_porcupine.so)
      6505==    by 0x4063823: pv_mel_filter_bank_compute (in /root/Arun/Porcupine-master/lib/linux/i386/libpv_porcupine.so)
      ==6505==    by 0x40619C4: pv_porcupine_multiple_keywords_process (in /root/Arun/Porcupine-master/lib/linux/i386/libpv_porcupine.so)
      6505==    by 0x404AB38: start_thread (in /lib/libpthread-2.12.so)
      6505==    by 0x428ED6D: clone (in /lib/libc-2.12.so)
      ######
   

Please note, the above crash doesn’t occur, when the second thread creates its porcupine handle after the first thread finishes the execution.
Could you please have a look at this issue and let us know the root cause and resolution?

Regards,
Arun

TL;DR :

1) Is there any way to make Porcupine log more useful debugging info to LogCat in Android?

2) Does Porcupine internally detects beginning and end of speech in the audio data, similar to how PocketSphinx or Snowboy do this? And if it does, is there any way to get these states?

Issue description:

I'm currently integrating Porcupine into my Android app and the issue that I'm facing is that it doesn't detect anything.

Porcupine happily reports to logcat that it got initialized correctly with proper parameters:

06-19 19:08:21.378 11391-11808/ I/PORCUPINE: �[32m [INFO] model file path: /storage/emulated/0/Android/data/com.example.app/files/sync/porcupine/porcupine_params.pv�[0m
    �[32m [INFO] number of keywords: 1�[0m
    �[32m [INFO] keyword file path [0]: /storage/emulated/0/Android/data/com.example.app/files/sync/porcupine/models/alexa_android.ppn�[0m
    �[32m [INFO] sensitivity [0]: 0.500000�[0m

Sample rate and frame length were correctly set up in according to what porcupine requires using Porcupine#getFrameLength() and Porcupine#getSampleRate(). Porcupine does not show any errors about sample rate or frame length when I feed it audio data.

Keyword: Alexa (other available keywords)
OS: Android

PS: here's, to better illustrate idea from the 2nd quesiton, an example of how Snowboy reports if the any speech was detected in current audio frame (taken from here):

// Snowboy hotword detection.
int result = detector.RunDetection(audioData, audioData.length);

if (result == -2) {
   // post a higher CPU usage:
    // sendMessage(MsgEnum.MSG_VAD_NOSPEECH, null);
} else if (result == -1) {
    sendMessage(MsgEnum.MSG_ERROR, "Unknown Detection Error");
} else if (result == 0) {
     // post a higher CPU usage:
     // sendMessage(MsgEnum.MSG_VAD_SPEECH, null);
} else if (result > 0) {
     sendMessage(MsgEnum.MSG_ACTIVE, null);
     Log.i("Snowboy: ", "Hotword " + Integer.toString(result) + " detected!");
     player.start();
}

Thanks.

As of now, I want to integrate Porcupine with Java through JNI, as you know this means that the dll must be built specifically for JNI Integration (through JNI Wrappers)

JNIEXPORT void JNICALL Java_package_name_ClassName_methodName(JNIEnv* env, jobject thiz) {}

I have only found this wrappers in android's .so libs, but because I want to implement it in Java, I cannot implement it on a Windows PC. I am using an AMD64 arch. Is there any way I can build this on my own? If not, could you provide me the modified dll?

is the min frame length for porcupine 512??

using Python3.6 for testing

i have an audio file with a frame length of 256.. and when running porcupine.process on this it never picks up the hotword

the audio stream i am using is

ChunkID= b'RIFF'
TotalSize= 556
DataSize= 512
Format= b'WAVE'
SubChunk1ID= b'fmt '
SubChunk1Size= 16
AudioFormat= 1
NumChannels= 1
SampleRate= 16000
ByteRate= 32000
BlockAlign= 2
BitsPerSample= 16
SubChunk2ID= b'data'
SubChunk2Size= 512

Hi,

Can I initialise and use porcupine in a background service that listens for the wake word continuously?

Expected behavior

A generated ppn file with the request keyword

Actual behavior

[WARN] This version of optimizer cannot create keyword files for ios. Please contact [email protected].

Steps to reproduce the behavior

tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources/ -w "ok google" -p ios -o ~/

Running on macOS High Sierra 10.13.4

I'm trying to implement this library in Android.
Here's my code:
onCreate()

copyPorcupineConfigFiles(this);
        String keywordFilePath = new File(this.getFilesDir(), "francesca.ppn")
                .getAbsolutePath();
        String modelFilePath = new File(this.getFilesDir(), "params.pv").getAbsolutePath();
        try {
            manager = new PorcupineManager(modelFilePath, keywordFilePath, sensitivity, new KeywordCallback() {
                @Override
                public void run(int keyword_index) {
                    Toast.makeText(MainActivity.this, "Detected!", Toast.LENGTH_LONG).show();
                }
            });
            manager.start();
        } catch (PorcupineManagerException e) {
            e.printStackTrace();
        }

copyPorcupineConfigFiles()

private static void copyPorcupineConfigFiles(Context context) {
        int[] resIds = {R.raw.francesca, R.raw.params};
        Resources resources = context.getResources();
        for (int resId : resIds) {
            String filename = resources.getResourceEntryName(resId);
            String fileExtension = resId == R.raw.params ? ".pv" : ".ppn";
            InputStream is = null;
            OutputStream os = null;
            try {
                is = new BufferedInputStream(resources.openRawResource(resId),
                        256);
                os = new BufferedOutputStream(context.openFileOutput(filename + fileExtension,
                        Context.MODE_PRIVATE), 256);
                int r;
                while ((r = is.read()) != -1) {
                    os.write(r);
                }
                os.flush();
            } catch (IOException e) {
                Toast.makeText(context, "Error!", Toast.LENGTH_SHORT).show();
            } finally {
                try {
                    if (is != null) {
                        is.close();
                    }
                    if (os != null) {
                        os.close();
                    }
                } catch (IOException e) {
                    Toast.makeText(context, "Error!", Toast.LENGTH_SHORT).show();
                }
            }
        }
    }

keywordFilePath is /data/user/0/com.package.name/files/francesca.ppn
modelFilePath is /data/user/0/com.package.name/files/params.pv

Both paths are correct.

When I run, I get this Porcupine error:

07-30 17:36:51.343 18194-18194/? I/PORCUPINE: [ERROR] loading parameter file failed with 'IO_ERROR'
07-30 17:36:51.345 18194-18194/? W/System.err: ai.picovoice.porcupinemanager.PorcupineManagerException: ai.picovoice.porcupine.PorcupineException: java.io.IOException: Initialization of Porcupine failed.

Installing on Raspberry PI Zero W with Raspbian, demo runs for about 1-2 seconds and throws:

Traceback (most recent call last): File "demo/python/porcupine_demo.py", line 204, in <module> input_device_index=args.input_audio_device_index).run() File "demo/python/porcupine_demo.py", line 104, in run pcm = audio_stream.read(porcupine.frame_length) File "/usr/local/lib/python2.7/dist-packages/pyaudio.py", line 608, in read return pa.read_stream(self._stream, num_frames, exception_on_overflow) IOError: [Errno -9981] Input overflowed
Maybe I have to change the frame size? If so, where do I change parameters?

I tested using python demo/python/porcupine_demo.py --keyword_file_paths resources/keyword_files/alexa_raspberrypi.ppn --output_path ~/testporcupine.wav
it records the file for 2 seconds and stops. I can hear my voice fine in those 2 sec.

ai.picovoice.porcupinemanager.PorcupineManagerException: ai.picovoice.porcupine.PorcupineException: java.lang.IllegalArgumentException: Initialization of Porcupine failed.

String keywordFilePath = new File(this.getFilesDir(), filename + ".ppn")
                .getAbsolutePath();
        String modelFilePath = new File(this.getFilesDir(), "params.pv").getAbsolutePath();
        try {
            manager = new PorcupineManager(modelFilePath, keywordFilePath, sensitivity, new KeywordCallback() {
                @Override
                public void run(int keyword_index) {
                    Toast.makeText(MainActivity.this, "Detected!", Toast.LENGTH_LONG).show();
                }
            });
            manager.start();
        } catch (PorcupineManagerException e) {
            e.printStackTrace();
            Log.e("Porcupine",e.getMessage());
        }

I have manually checked if keywordFilePath and modelFilePath are correct, and they are. This also happens in your demo application.

We want to install Picovoice service in Asus Tinker Board (https://en.wikipedia.org/wiki/Asus_Tinker_Board)
We have arm71 machine.
But we've got an axception "Cannot autodetect the binary type. Please enter the path to the shared object using --library_path command line argument."
What library do we need ?
Can we run Picovoice service on this board at all ?

Thanks

Expected behavior

Use porcupine in C/C++ Android NDK native program.

Actual behavior

"undefined reference" to every porcupine function.

Steps to reproduce the behavior

Write a C program;
include porcupine C headers;
call API C functions;
link against libpv_porcupine.so found in android arm_v7a folder;
build with android NDK v15c.

Longer awake work doesn't work?

no detection of 3 words in a row...

Is there any way to detect 3 words in a row? please? can the model be expended a bit?

Expected behavior

python demo/python/porcupine_demo.py --keyword_file_paths resources/keyword_files/alexa_linux.ppn

Demo runs.

Actual behavior

Error:

Traceback (most recent call last):
  File "demo/python/porcupine_demo.py", line 210, in <module>
    input_device_index=args.input_audio_device_index).run()
  File "demo/python/porcupine_demo.py", line 95, in run
    sensitivities=[self._sensitivity] * num_keywords)
  File "demo/python/../../binding/python/porcupine.py", line 69, in __init__
    library = cdll.LoadLibrary(library_path)
  File "/usr/local/conda3/lib/python3.6/ctypes/__init__.py", line 426, in LoadLibrary
    return self._dlltype(name)
  File "/usr/local/conda3/lib/python3.6/ctypes/__init__.py", line 348, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: demo/python/../../lib/linux/x86_64/libpv_porcupine.so: failed to map segment from shared object

Steps to reproduce the behavior

Environment:

ChromeOS on Acer CXI2
Linux localhost 3.14.0 #1 SMP PREEMPT Fri Jun 22 17:20:26 PDT 2018 x86_64 Intel(R) Core(TM) i7-5500U CPU @ 2.40GHz GenuineIntel GNU/Linux

Steps to reproduce:
Commented out soundfile dependency because libsndfile is not installed in the ChromeOS linux.
Start terminal with ChromeOS (Ctrl + Alt + T)
Clone Porcupine repo.
Installed portaudio, pyaudio.
Run python: python demo/python/porcupine_demo.py --keyword_file_paths resources/keyword_files/alexa_linux.ppn

I know ChromeOS Linux is not the typical use case. Looking for any guidance on where this error could stem from since failed to map segment from shared object is not specific for me.

Expected behavior

Actual behavior

Steps to reproduce the behavior

Hi I am using Porcupine on Linux (Ubuntu 64-bit) for making a wake-up word "dayo". But I could not do that. I am getting error mesage like [ERROR] could not find the pronunciation for 'dayo'. If this is not a typo please contact [email protected].
Can you please add the word to Porcupine to make the word as wakeup keyword or can you suggest a procedure to create a custom wakeup word like I used above(dayo)?
The 'dayo' is pronounced as: https://www.howtopronounce.com/dayo/

Hi,

Trying to build my custom wake word , for mac

I am running the following command :
./pv_porcupine_optimizer -r resources -w google -p mac -o ~/
./pv_porcupine_optimizer -r resources -w yellow -p mac -o ~/
./pv_porcupine_optimizer -r resources -w 'action item' -p mac -o ~/
./pv_porcupine_optimizer -r resources -w 'take note' -p mac -o ~/

getting the following:
[ERROR] Could not find a pronunciation for 'google'. If this is not a typo please contact [email protected].

Steps to reproduce the behavior

./pv_porcupine_optimizer -r resources -w google -p mac -o ~/

Hello!
I tried to use the optimizer with paired keywords like "ok" + the name of a french brand but without success. Typically, "OK " as a wake-word.

Is there any way to extend the pronunciation directory ? For example by providing audio data samples of french users saying this brand name?

Best,
Denis

I designed a logo for Porcupine. In this logo you can see simplicity and elegance more clearly. I used the letter P to indicate the real purpose. Even when we reduce it to size, it is so simple that there will be no distortion. I think it's a logo designed for you, I hope you like it.

Can you add the keyword "lifetouch", thank you in advance. Great software.

Hi, and thanks for your great product

My problem is that When i try to run
tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources/ -p mac -o . -w "hey rimon"
I will get the following error:
[ERROR] could not find the pronunciation for 'hey rimon'
When I've read the docs, it was mentioned that only the words in english are supported.
I wanted to know is there anyway that I can add the pronunciation to the dictionary so I will be able to get my costume keyword wake up model?
Thanks

Expected behavior

When embedded for iOS project, it should be compilable for both deploying to device and simulator.

Actual behavior

It'll only compile if the target is to actual device.

Steps to reproduce the behavior

Compile the demo project on simulator, it will fail due to the binary was built only for device architecture.

I think it's probably worth it to provide a universal binary for iOS, otherwise it'll mean that adding this library to the project will cause that the project no longer compilable for simulator. The barrier of entry to adopt this library will be really high.

"Below is a quick demonstration of how to construct an instance of it to detect multiple keywords concurrently"

I'm a beginner in code, so I'm not sure how this works. Do I edit the python binding code? and when I call it in my command prompt do I still need the keyword_file_path header?

I was looking at the wakeword-benchmark utility and I was wondering what data was used to train the alexa_<PLATFORM>.ppn models.

You might consider reducing size to speed up downloads

Hello,

In Python, we can do multiple wake word detection as documented below:

library_path = ... # Path to Porcupine's C library available under lib/${SYSTEM}/${MACHINE}/
model_file_path = ... # It is available at lib/common/porcupine_params.pv
keyword_file_paths = ['path/to/keyword/1', 'path/to/keyword/2', ...]
sensitivities = [0.5, 0.4, ...]
handle = Porcupine(library_path, model_file_path, keyword_file_paths=keyword_file_paths, sensitivities=sensitivities)

How do we do the same on Android? Can someone help?

I tried to run the demo command like this:
python demo/python/porcupine_demo.py --keyword_file_paths resources/keyword_files/blueberry_windows.ppn

But i got this error:

Traceback (most recent call last):
  File "demo/python/porcupine_demo.py", line 204, in <module>
    input_device_index=args.input_audio_device_index).run()
  File "demo/python/porcupine_demo.py", line 92, in run
    sensitivities=[self._sensitivity] * num_keywords)
  File "demo/python\../../binding/python\porcupine.py", line 69, in __init__
    library = cdll.LoadLibrary(library_path)
  File "C:\Python\lib\ctypes\__init__.py", line 426, in LoadLibrary
    return self._dlltype(name)
  File "C:\Python\lib\ctypes\__init__.py", line 348, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: [WinError 126] The specified module could not be found

What can i do to fix this problem

• Fredrik

Expected behavior

Archive without failing.

Actual behavior

When we Archive the build (iOS demo), it fails with the following error:
ld: bitcode bundle could not be generated because '/Users/userXXXX/Code/vivaa/native/Scribe/ios/Porcupine/watchos/libpv_porcupine.a(pv_porcupine.o)' was built without full bitcode. All object files and libraries for bitcode must be generated from Xcode Archive or Install build for architecture armv7k

Steps to reproduce the behavior

1. Download Porcupine iOS Demo
2. Product > Archive
3. See error

Please include the operating system and CPU architecture. When applicable, provide relevant audio data.
Happens with iOS and watchOS demos.

Hello,

First of all, thank you for this engine, it's awesome!

I've tried to create a small wake up sentence using the optimizer command line too, like the famous "ok google" but it seems not to be supported.

Command:

tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources/ -w "ok test" -p mac -o ~/

Output:
[ERROR] Could not find a pronunciation for 'ok test'. If this is not a typo please contact [email protected].

On my real use case the trigger command will be "Ok + 2 small words"

Is this the expected behavior of Porcupine? Which means Porcupine is designed exclusively for single words detection? If so, any workaround?
Or did I miss something on the optimizer tool?

Thank you.

Is there a way to get the detected audio in the callback?
It would be really nice to have the callback include the audio frames so e.g. cloud-based wake word verification for the Alexa Voice Service can be used.

I'm running Visual Studio 2017 on a Windows 10 machine.

I've tried this with both 32- and 64-bit projects.

The result I get is similar in each case:
4>..\..\porcupine\lib\libpv_porcupine.dll : fatal error LNK1107: invalid or corrupt file: cannot read at 0x430 (64-bit)
1>..\..\porcupine\lib\libpv_porcupine.dll : fatal error LNK1107: invalid or corrupt file: cannot read at 0x3F8 (32-bit)

I also tried previous committed versions with no luck.

Expected behavior

Actual behavior

I used this tutorial https://youtu.be/3z7LBW_Rl9c.

I have Ubuntu 16.04, Intel Core i3 7th Gen.
As a result of running service I got error.
Error log:
Traceback (most recent call last):
File "demo/python/porcupine_demo.py", line 207, in
input_device_index=args.input_audio_device_index).run()
File "demo/python/porcupine_demo.py", line 92, in run
sensitivities=[self._sensitivity] * num_keywords)
File "demo/python/../../binding/python/porcupine.py", line 114, in init
byref(self._handle))
File "/home/scale/.local/lib/python2.7/site-packages/enum.py", line 199, in init
raise EnumBadKeyError(key)
enum.EnumBadKeyError: Enumeration keys must be strings: 0

Can I customize a Chinese wakeup keyword?

Expected behavior

Actual behavior

Steps to reproduce the behavior

Please include the operating system and CPU architecture. When applicable, provide relevant audio data.

Hi, I'm making some prototype and try to make my keyword like below.
But it's not work with pronunciation error.
Can you suggest the way to resolve this problem or add a keyword "ariot" (not "a riot")??
Please advice about it.

Thanks.

C:\Users\1\Porcupine-master\Porcupine-master>tools\optimizer\windows\i686\pv_porcupine_optimizer.exe -r resources\ -w "ariot" -p windows -o .
[ERROR] could not find the pronunciation for 'ariot'. If this is not a typo please contact [email protected].

Expected behavior

I'm trying to integrate Porcupine into an xcode project as per Porcupine/demo/ios/PorcupineDemo.
PorcupineDemo runs perfectly, so I copied the necessary components and settings of the demo to my project, but alas, it does not compile.

Actual behavior

Ld [~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Products/Debug-iphonesimulator/Porcupine\ Test.app/Porcupine\ Test normal x86_64
    cd "[~]/Library/Autosave Information/Porcupine Test"

    export IPHONEOS_DEPLOYMENT_TARGET=11.3

    export PATH="/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneSimulator.platform/Developer/usr/bin:/Applications/Xcode.app/Contents/Developer/usr/bin:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin"

    /Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang -arch x86_64 -isysroot /Applications/Xcode.app/Contents/Developer/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator11.3.sdk -L[~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Products/Debug-iphonesimulator -L[~]/Library/Autosave\ Information/Porcupine\ Test/lib/ios -F[~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Products/Debug-iphonesimulator -filelist[~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Intermediates.noindex/Porcupine\ Test.build/Debug-iphonesimulator/Porcupine\ Test.build/Objects-normal/x86_64/Porcupine\ Test.LinkFileList -Xlinker -rpath -Xlinker @executable_path/Frameworks -mios-simulator-version-min=11.3 -dead_strip -Xlinker -object_path_lto -Xlinker [~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Intermediates.noindex/Porcupine\ Test.build/Debug-iphonesimulator/Porcupine\ Test.build/Objects-normal/x86_64/Porcupine\ Test_lto.o -Xlinker -export_dynamic -Xlinker -no_deduplicate -Xlinker -objc_abi_version -Xlinker 2 -fobjc-link-runtime -L/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/swift/iphonesimulator -Xlinker -add_ast_path -Xlinker [~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Intermediates.noindex/Porcupine\ Test.build/Debug-iphonesimulator/Porcupine\ Test.build/Objects-normal/x86_64/Porcupine_Test.swiftmodule -Xlinker -sectcreate -Xlinker __TEXT -Xlinker __entitlements -Xlinker [~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Intermediates.noindex/Porcupine\ Test.build/Debug-iphonesimulator/Porcupine\ Test.build/Porcupine\ Test.app-Simulated.xcent -lpv_porcupine -Xlinker -dependency_info -Xlinker [~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Intermediates.noindex/Porcupine\ Test.build/Debug-iphonesimulator/Porcupine\ Test.build/Objects-normal/x86_64/Porcupine\ Test_dependency_info.dat -o [~]/Library/Developer/Xcode/DerivedData/Porcupine_Test-bauzordzwumanjdyzvkejvhihxcl/Build/Products/Debug-iphonesimulator/Porcupine\ Test.app/Porcupine\ Test

error: Invalid bitcode signature
clang: error: linker command failed with exit code 1 (use -v to see invocation)

Steps to reproduce the behavior

I) create new xcode project named 'Porcupine Test'
II) copy

1. Porcupine/include to ${PROJECT_DIR}/include

2. Porcupine/lib to ${PROJECT_DIR}/lib

3. Porcupine/binding/ios/PorcupineManager to ${PROJECT_DIR}/Porcupine Test/PorcupineManager

4. Porcupine/demo/ios/PorcupineDemo/module.modulemap to ${PROJECT_DIR}/Porcupine Test/Porcupine/module.modulemap

   a) edit module.modulemap relative paths to '../../include'

III) set Build Settings

1. Search Paths -> Library Search Paths to ${PROJECT_DIR}/lib/ios

2. Swift Compiler - Search Paths -> Import Paths to ${PROJECT_DIR}/Porcupine\ Test/Porcupine

IV) add ${PROJECT_DIR}/lib/ios/libpv_porcupine.a to Build Phases -> Link Binary With Libraries

V) build project

xcode 9.3 (9E145)
macos 10.13.4
1.6 GHz Intel Core i5

I tried to compile the iOS demo, but i am getting errors while compiling
Error is:
Undefined symbols for architecture x86_64:
"_pv_porcupine_delete", referenced from:
PorcupineDemo.PorcupineManager.stop() -> () in PorcupineManager.o
"_pv_sample_rate", referenced from:
PorcupineDemo.PorcupineManager.start() throws -> () in PorcupineManager.o
"_pv_porcupine_frame_length", referenced from:
PorcupineDemo.PorcupineManager.start() throws -> () in PorcupineManager.o
"_pv_porcupine_multiple_keywords_init", referenced from:
PorcupineDemo.PorcupineManager.start() throws -> () in PorcupineManager.o
"_pv_porcupine_multiple_keywords_process", referenced from:
closure #1 (Swift.UnsafeMutableRawPointer?, Swift.OpaquePointer, Swift.UnsafeMutablePointer<__C.AudioQueueBuffer>, Swift.UnsafePointer<__C.AudioTimeStamp>, Swift.UInt32, Swift.UnsafePointer<__C.AudioStreamPacketDescription>?) -> () in variable initialization expression of PorcupineDemo.PorcupineManager.(audioCallback in _BEC5063E4C49B2A2811E8F4D93649EFF) : @convention(c) (Swift.UnsafeMutableRawPointer?, Swift.OpaquePointer, Swift.UnsafeMutablePointer<__C.AudioQueueBuffer>, Swift.UnsafePointer<__C.AudioTimeStamp>, Swift.UInt32, Swift.UnsafePointer<__C.AudioStreamPacketDescription>?) -> () in PorcupineManager.o
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)

Can you please help me to compile the demo build

so we can't built models for Raspberry Pi ? we can only buy one for the RPi ?

Hello
Thank you for the great work!

I have the following problem. When I run the following command:

tools\optimizer\windows\i686\pv_porcupine_optimizer -r resources/ -w cerner -p windows -o .

I get the following error:

[ERROR] Could not find the pronunciation for 'cerner'.

Is there a method to add this word to your vocabulary using the optimizer tool?

Thanks
Emins

I tried to integrate it into python but i got this error:

Traceback (most recent call last):
  File "SpeechRecognition.py", line 14, in <module>
    handle = Porcupine(library_path, model_file_path, keyword_file_paths=keyword_file_paths, sensitivities=sensitivities)
NameError: name 'Porcupine' is not defined

And this is my code:

# WAKE WORD
library_path = ['WakeWord/lib/${SYSTEM}/${MACHINE}/']
model_file_path = ['WakeWord/lib/common/porcupine_params.pv']
keyword_file_paths = ['WakeWord/resources/keyword_files/alexa_windows.ppn']
sensitivities = [0.5, 0.4]
handle = Porcupine(library_path, model_file_path, keyword_file_paths=keyword_file_paths, sensitivities=sensitivities)
# WAKE WORD END

def get_next_audio_frame():
    pass

while True:
    pcm = get_next_audio_frame()
    keyword_index = handle.process(pcm)
    if keyword_index >= 0:
        # detection event logic/callback
        pass

(Just the default from github)

• Fredrik :)

i tried creating a new wake word using the following command
tools/optimizer/mac/x86_64/pv_porcupine_optimizer -r resources -w “hello world” -p mac -o ~/Users/mahmoud/Desktop
when I use one word, example hello it gives the pronunciation error "tried many other words", when i use two or more i get
[ERROR] invalid arguments
the test file is working fine, but i can't add any other words.

any ideas?

OS: Mac OS high sierra (10.13.4)