Error if text > 200 char

[antoniofagardo]

I am currently running version 2.2.0 and it seems like if the text we are trying to convert is greater than 200 char, the conversion fails with the following message:

Traceback (most recent call last):
  File "main.py", line 11, in <module>
    myobj.save("audiobook.mp3")
  File "/usr/local/lib/python3.8/dist-packages/gtts/tts.py", line 311, in save
    self.write_to_fp(f)
  File "/usr/local/lib/python3.8/dist-packages/gtts/tts.py", line 293, in write_to_fp
    raise gTTSError(tts=self, response=r)
gtts.tts.gTTSError: 200 (OK) from TTS API. Probable cause: Unknown

Sample code to reproduce the problem:

from gtts import gTTS

finalChar = 201
language = 'en'
finalText = ""
for i in range(30):
    finalText += "\n0123456789"
print(finalText[:finalChar])

myobj = gTTS(text=finalText[:finalChar], lang=language, slow=False)
myobj.save("audiobook.mp3")

if you change back finalChar to 200, the request works and the file is written on disk.

[pndurette]

@antoniofagardo Thanks for the report, let me take a look.

[pndurette]

Yeah, there's a weird bug with how requests are prepared I think. It sends way too much data to the API. Investigating more.

[pndurette]

Ah found it—
_package_rpc() takes the entire text submitted and not a part. So say it splits the text in 10 parts, the entire text is sent 10 times.

So the error was indeed that the API was getting too many characters (I guess it's 200 now!) and not returning any audio. Which was an error that gTTSError couldn't interpret with the info it had, hence "Unknown".

Fixing and will push a new version.

[pndurette]

Pushed gTTS 2.2.1 fixing the problem! Thanks for the bug report!

[antoniofagardo]

Thank you for the quick fix!

[KelvinKramp]

Ive got the same problem, but i think due to a different reason

Traceback (most recent call last):
File "/Users/mackliniek/Tests/morfine_protocol.py", line 32, in
welcome()
File "/Users/mackliniek/Tests/morfine_protocol.py", line 30, in welcome
SR.text_speech(morfine_conversation[0])
File "/Users/mackliniek/Tests/TTS_SST.py", line 28, in text_speech
speech.save('speech.mp3')
File "/Users/mackliniek/dashboard/venv/lib/python3.9/site-packages/gtts/tts.py", line 312, in save
self.write_to_fp(f)
File "/Users/mackliniek/dashboard/venv/lib/python3.9/site-packages/gtts/tts.py", line 294, in write_to_fp
raise gTTSError(tts=self, response=r)
gtts.tts.gTTSError: 200 (OK) from TTS API. Probable cause: Unknown

Is there any fix? I have installed gTTS 2.2.1.

[AngelosDouglas]

Mesmo problema aqui

[KelvinKramp]

I found the problem in my case, the translation accent was set to "en-ca" representing english canadian. After some debugging (sweat over my body, tears dropping down my head, axillary region all wet etc. :-)) I found it. Maybe you should check what happens if you change the speaking accent to English with "en", I also removed slow=False. Check it out and let me know if it works out.

[srugai]

Unbelievable, the code below was working like a charm for two months now (gTTs 2.2.1). Started to get the error commented here. Tried KelvinKramp approach and it is working again. I mean the error stopped but lang="en" and lang="en-au" produces a different result right?

def text_to_speech():
tts = gtts.gTTS(T.get("1.0", tk.END), lang='en-au') # 'en-au': 'English (Australia)'
mp3_file = "_temp.mp3"
tts.save(mp3_file)
music = AudioSegment.from_mp3(mp3_file)
print("Playing mp3 file...")
play(music)

[pndurette]

@KelvinKramp and @srugai

Hi there, you're right, Google has removed upstream almost of the <Lang>-<geo> language coded that used to work (and yeah produce different accents).

I've removed them, see #267, but since they've been part of the library for years, I'm adding a compatibility fallback (eg "en-CA" falls back to "en").

[NikithaNimbalkar]

@KelvinKramp and @srugai

Hi there, you're right, Google has removed upstream almost of the <Lang>-<geo> language coded that used to work (and yeah produce different accents).

I've removed them, see #267, but since they've been part of the library for years, I'm adding a compatibility fallback (eg "en-CA" falls back to "en").

What to do if we actually want the accents...

[evinracher]

I was having the same issue. My code was:

from gtts import gTTS tts = gTTS('Hola mundo. Estamos convirtiendo texto a voz con Python.', lang='es-us')

I changed es-us to es and it worked

[erichsu0527]

Since zh is not a language, Chinese does not work at all.

from gtts import gTTS
tts = gTTS("说中文", lang="zh-CN")
tts.save("helloch.mp3")
tts = gTTS("說中文", lang="zh-TW")
tts.save("helloch.mp3")

Get "gtts.tts.gTTSError: 200 (OK) from TTS API. Probable cause: Unknown" on both tts.save()

[zesto01]

I am having the same issue with erichsu0527. Are there any workarounds?

[joeminicucci]

Same problem here as @zesto01 and @erichsu0527.
@pndurette I was having trouble finding the root cause debugging the code, would you be able to take a look? Would much appreciate your trouble as its effecting many users for an open source downstream I maintain.

Input:

tts = gTTS('的', lang='zh-CN', lang_check=True)
tts.save(self.path)

Trace:

  File "C:\Users\User1\AppData\Roaming\Anki2\addons21\chinese\tts.py", line 50, in download
    self.func()
  File "C:\Users\User1\AppData\Roaming\Anki2\addons21\chinese\tts.py", line 56, in get_google
    tts.save(self.path)
  File "C:\Users\User1\AppData\Roaming\Anki2\addons21\venv\lib\site-packages\gtts\tts.py", line 312, in save
    self.write_to_fp(f)
  File "C:\Users\User1\AppData\Roaming\Anki2\addons21\venv\lib\site-packages\gtts\tts.py", line 294, in write_to_fp
    raise gTTSError(tts=self, response=r)

gtts.tts.gTTSError: 200 (OK) from TTS API. Probable cause: Unknown

[pndurette]

@erichsu0527, @zesto01, @joeminicucci

I was able to run all of your code samples without any issues. Have you updated? There was an upstream change with languages, which was fixed in 2.2.2

[pndurette]

@KelvinKramp, @srugai & @NikithaNimbalkar:

The accents are back, just obtained differently!

I've added updated documentation on how to obtain local accents with the combination of the right language code (lang) and top-level domain (tld)! See: https://gtts.readthedocs.io/en/latest/module.html#localized-accents

[joeminicucci]

@pndurette I tested that code 5 days ago on 2.2.2. Let me try again.

[joeminicucci]

@pndurette Following the new tld convention (doesn't apply since Im using zh-CN and zh-TW only), Im still reliably reproducing the error.

[pndurette]

@pndurette Following the new tld convention (doesn't apply since Im using zh-CN and zh-TW only), Im still reliably reproducing the error.

@joeminicucci Ah damn, let's look. Do you mind opening a new issue with your problem? It's really getting difficult to follow exactly the issues everyone is experiencing as the current GitHub issue has gone all over the place.

[zesto01]

@joeminicucci @pndurette
I tried using 2.2.2 and lang="zh-CN", it is now working. though parameter slow=True, does no longer work.

[agnplu]

@pndurette
Will other local accents be available too? I only see zh-CN and zh-TW now, and I'm mostly interested in en-US.

[pndurette]

@agnplu Please open new a Discussion for questions—but for "en-US", see here. You should be able to use lang='en' with tld='com' (which is the default)