Issue simply getting azure_tts-i2s example working #1297
-
|
Hello, I have been struggling to get the azure tts example working. I tried the example as it was (with api key, etc filled in), but that didn't work. I tried to narrow down the issue by just using a CSV output instead of the I2S output, but still it "fails to write". Here's my code and debug output. Any help would be really appreciated! I've also tried a simple SineWaveGenerator with I2S and it works perfectly/produces output. I'm using a esp32-eye dev board and an adafruit max98.... breakout. Already have #include <Arduino.h>
#include "AudioTools.h"
//AudioInfo info(44100, 2, 16);
String speechKey = "..."; // deploy a Speech Service in Azure and get both the key and the region. info here: https://azure.microsoft.com/en-us/products/cognitive-services/text-to-speech/
String spechregion = "eastus";
String voice = "en-US-JennyNeural"; // for the next 3 settings, chose from: https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support?tabs=stt#prebuilt-neural-voices
String gender = "Female";
String language = "en-US";
String url_str = "https://" + spechregion + ".tts.speech.microsoft.com/cognitiveservices/v1";
String msg = "This is a demonstration of Phil Schatzmann's AudioTools integrating with the Azure Speech Service. Hope you like it.";
const char* ssid = "iPhone (65)";
const char* password = "testpass";
WiFiClientSecure client_spk;
//URLStream AzureURLStream("iPhone (65)", "testpass");
URLStream AzureURLStream(client_spk);
//I2SStream i2s; // or I2SStream
CsvOutput<int32_t> csvStream(Serial);
WAVDecoder decoder;
EncodedAudioStream out(&csvStream, &decoder); // Decoder stream
StreamCopy copier(out, AzureURLStream); // copy in to out
void setup(){
Serial.begin(115200);
AudioLogger::instance().begin(Serial, AudioLogger::Debug);
WiFi.begin(ssid, password);
while (WiFi.status() != WL_CONNECTED) {
delay(1500);
Serial.println("Connecting to WiFi...");
}
Serial.println("Connected to WiFi");
client_spk.setInsecure();
String ssml = "<speak version='1.0' xml:lang='" + language + "'><voice xml:lang='" + language + "' xml:gender='" + gender + "' name='" + voice + "'>" + msg + "</voice></speak>";
AzureURLStream.addRequestHeader("Ocp-Apim-Subscription-Key", speechKey.c_str());
AzureURLStream.addRequestHeader("X-Microsoft-OutputFormat", "riff-16khz-16bit-mono-pcm"); // if you change this, change the settings for i2s and the decoder
AzureURLStream.addRequestHeader(USER_AGENT, String("Arduino with Audiotools version:" + String(AUDIOTOOLS_VERSION)).c_str());
AzureURLStream.begin(url_str.c_str(), "audio/wav", POST, "application/ssml+xml", ssml.c_str());
csvStream.begin();
}
void loop(){
copier.copy();
} |
Beta Was this translation helpful? Give feedback.
Answered by
pschatzmann
Feb 6, 2024
Replies: 3 comments 5 replies
-
|
out is missing a begin() call, so it is not open... |
Beta Was this translation helpful? Give feedback.
2 replies
-
|
That fixed it! Thanks so much! Now it produces perfect audio for like 1 second, then the esp crashes. Do you have any experience with this issue? I can make a pr with that line added to the example. #include <Arduino.h>
/**
* @file streams-azure_tts-i2s.ino
* @author Kevin Saye
* @copyright GPLv3
*
*/
#include "AudioTools.h"
//AudioInfo info(44100, 2, 16);
String speechKey = "..."; // deploy a Speech Service in Azure and get both the key and the region. info here: https://azure.microsoft.com/en-us/products/cognitive-services/text-to-speech/
String spechregion = "eastus";
String voice = "en-US-JennyNeural"; // for the next 3 settings, chose from: https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support?tabs=stt#prebuilt-neural-voices
String gender = "Female";
String language = "en-US";
String url_str = "https://" + spechregion + ".tts.speech.microsoft.com/cognitiveservices/v1";
String msg = "This is a demonstration of Phil Schatzmann's AudioTools integrating with the Azure Speech Service. Hope you like it.";
const char* ssid = "iPhone (65)";
const char* password = "testpass";
WiFiClientSecure client_spk;
//URLStream AzureURLStream("iPhone (65)", "testpass");
URLStream AzureURLStream(client_spk);
I2SStream i2s; // or I2SStream
//CsvOutput<int32_t> csvStream(Serial);
WAVDecoder decoder;
EncodedAudioStream out(&i2s, &decoder); // Decoder stream
StreamCopy copier(out, AzureURLStream); // copy in to out
void setup(){
Serial.begin(115200);
//AudioLogger::instance().begin(Serial, AudioLogger::Debug);
WiFi.begin(ssid, password);
while (WiFi.status() != WL_CONNECTED) {
delay(1500);
Serial.println("Connecting to WiFi...");
}
Serial.println("Connected to WiFi");
client_spk.setInsecure();
// setup i2s
auto config = i2s.defaultConfig(TX_MODE);
//config.
config.sample_rate = 16000;
config.bits_per_sample = 16;
config.channels = 1;
config.i2s_format = I2S_LSB_FORMAT;
//config.copyFrom(info);
config.pin_ws = 2; //LCK
config.pin_bck = 12; //BCK
config.pin_data = 19; //DIN
i2s.begin(config);
// Source: https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/get-started-text-to-speech?tabs=windows%2Cterminal&pivots=programming-language-rest
String ssml = "<speak version='1.0' xml:lang='" + language + "'><voice xml:lang='" + language + "' xml:gender='" + gender + "' name='" + voice + "'>" + msg + "</voice></speak>";
AzureURLStream.addRequestHeader("Ocp-Apim-Subscription-Key", speechKey.c_str());
AzureURLStream.addRequestHeader("X-Microsoft-OutputFormat", "riff-16khz-16bit-mono-pcm"); // if you change this, change the settings for i2s and the decoder
AzureURLStream.addRequestHeader(USER_AGENT, String("Arduino with Audiotools version:" + String(AUDIOTOOLS_VERSION)).c_str());
AzureURLStream.begin(url_str.c_str(), "audio/wav", POST, "application/ssml+xml", ssml.c_str());
//while(copier.available())
out.begin();
//csvStream.begin();
}
void loop(){
copier.copy();
} |
Beta Was this translation helpful? Give feedback.
-
|
Better backtrace: |
Beta Was this translation helpful? Give feedback.
Answer selected by
pschatzmann
-
|
What would be the related debug log ? I don't recall having seen any chunked result. Maybe there is an issue related to that. I propose that you try to change the protocol to HTTP/1.0 What is the reason for your strange and complicated https logic ? |
Beta Was this translation helpful? Give feedback.
2 replies
-
|
I have the http logic so I can do other stuff with the wifi connection outside the URLStream. For testing I've tried just doing: And the error persists. Attached is a log of with info level logs. If I do debug level, it plays slow and the error does not occur. Notably in the log I sent here there is no coredump but the audio stops playing nonetheless prematurely. Do you maybe have a esp32 on hand you could try to reproduce this on? Sorry this error is so strange. I'm not sure how to change the http protocol with the |
Beta Was this translation helpful? Give feedback.
-
|
Unfortunately I do not have any active Azure account. |
Beta Was this translation helpful? Give feedback.
-
|
Ok, if I manually set (Edit: Last question-when it starts playing audio there's about a half second of static where the audio is unintelligible, then the rest is fine. Is there a way for me to fix this? |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
With the last commit, the setProtocol should be available. Looking at the log, the issue seems to come from some trailing reply parameters. You could also try to call setAutoCreateLines(false), so that the system does not try to save these values which seem to be much too long. You would really need to analyse where the noise is coming from. Maybe it's a feature of your audio chip or your arduino (esp32?) core version. You could try to fill the i2s buffer with silence first by calling a couple of i2s.writeSilence(1024) or set the dac/amplifier chip to active only when you receive the data. I think usually users using a MAX98357A have this issue... |
Beta Was this translation helpful? Give feedback.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
out is missing a begin() call, so it is not open...