[Question] How to use this project?

[NightMachinery]

Prescript: I posted this on https://discourse.mozilla.org/t/using-the-tts-on-a-macbook-pro/44140, but I got no response, so I'm posting it here.

My eyes have been giving me trouble again, and I am looking to see if there are any usable open-source TTS engines. I listened to the samples at https://github.com/mozilla/TTS/wiki/Released-Models, and I like what I’ve seen. Is it possible to use these pretrained model on a GPUless Macbook Pro to create audiobooks? I am ok with a speed of even 0.1. If it is possible, how do I get it to work? I want to ultimately have a shell function that takes a txt file and gives me an audio file. Googling led me to this ‘demo server’ thingy at https://github.com/mozilla/TTS/tree/distributed-dev/server, which is a 404 currently.

[tugstugi]

@NightMachinary I haven't also found a guide how to synthesize with this repo.

Maybe you can try other open source impelementations. Here are some notebooks which are executable on colab:

• NVidia Tacotron2 + Waveglow: https://colab.research.google.com/github/tugstugi/dl-colab-notebooks/blob/master/notebooks/NVidia_Tacotron2_Waveglow.ipynb
• fatchord Tacotron+WaveRNN: https://colab.research.google.com/github/tugstugi/dl-colab-notebooks/blob/master/notebooks/fatchordWaveRNN.ipynb

[tugstugi]

@NightMachinary here is a Colab notebook for Mozilla TTS + WaveRNN: https://colab.research.google.com/github/tugstugi/dl-colab-notebooks/blob/master/notebooks/Mozilla_TTS_WaveRNN.ipynb

It will automatically checkout the correct commit/branch, download the pretrained models and synthesize an audio. You have to change the SENTENCE variable in the notebook.

[nmstoker]

Take a look here at the demo server: https://github.com/mozilla/TTS/tree/master/server (I think you had the wrong location before)

If you read the code a bit, it shouldn't be too hard to figure out how call the API for it from a script that feeds it sentences from your audiobook text and then append the audio returned into a single file.

[thiscris]

If I want to have this as a TTS on my system as to apply it to Firefox narrate function in reader mode, what do I need to do?

[MikeyBeez]

It just doesn't work. Neither does deepspeech. Use festival and SpeechRecognition.

[reuben]

I've made a simple packaging option and included a model in the wiki: https://github.com/mozilla/TTS/wiki/Released-Models#simple-packaging---self-contained-package-that-runs-an-http-api-for-a-pre-trained-tts-model

Instructions are in the README.md file inside. Would appreciate any feedback.

[marimeireles]

a huge improvement! thanks @reuben :)

[IveJ]

Thanks for your hard work.

On Thu, Nov 28, 2019, 20:53 Mariana Meireles notifications@github.com wrote:

a huge improvement! thanks @reuben https://github.com/reuben :)

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/mozilla/TTS/issues/272?email_source=notifications&email_token=AEYAML57P64NW3OCIVN7LILQV7EMVA5CNFSM4IS62UUKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEFMVKXI#issuecomment-559502685, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEYAMLZHLNHRHETY7255TRDQV7EMVANCNFSM4IS62UUA .

[nayanhalder]

hi @reuben i have done the following steps.

1. Create a fresh virtual environment with Python 3.7
2. sudo apt-get install espeak libsndfile1
3. pip install https://github.com/reuben/TTS/releases/download/t2-ljspeech-mold/TTS-0.0.1+b6b513f-py3-none-any.whl

i am getting many errors while intallation of scipy in step 3.

Can you help me please. where i am getting wrong.

following error i am getting.

error: Command "gcc -pthread -B /home/nayan/anaconda3/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/nayan/tmp/tts-venv/lib/python3.7/site-packages/numpy/core/include -I/home/nayan/tmp/tts-venv/lib/python3.7/site-packages/numpy/core/include -I/home/nayan/anaconda3/include/python3.7m -c scipy/cluster/_hierarchy.c -o build/temp.linux-x86_64-3.7/scipy/cluster/_hierarchy.o -MMD -MF build/temp.linux-x86_64-3.7/scipy/cluster/_hierarchy.o.d" failed with exit status 1

Rolling back uninstall of scipy Moving to /home/nayan/tmp/tts-venv/lib/python3.7/site-packages/scipy-1.3.3.dist-info/ from /home/nayan/tmp/tts-venv/lib/python3.7/site-packages/~cipy-1.3.3.dist-info Moving to /home/nayan/tmp/tts-venv/lib/python3.7/site-packages/scipy/ from /home/nayan/tmp/tts-venv/lib/python3.7/site-packages/~cipy ERROR: Command errored out with exit status 1: /home/nayan/tmp/tts-venv/bin/python3 -u -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/tmp/pip-install-h08sh93d/scipy/setup.py'"'"'; file='"'"'/tmp/pip-install-h08sh93d/scipy/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(file);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' install --record /tmp/pip-record-o76kawn1/install-record.txt --single-version-externally-managed --compile --install-headers /home/nayan/tmp/tts-venv/include/site/python3.7/scipy Check the logs for full command output.

[reuben]

Use Python 3.6 as documented on the wiki page. Some of the dependencies predate Python 3.7, so if you're using that, it'll try to build them from source.

[nayanhalder]

Hi @ruben many many thanks for your help. i have created virtual environment with 3.6 and the flask app is working fine.

My requirement is to generate sppech frpm text via batch, so i do not require flask app. therefore i need to keep the input text file in somewhere /home/nayan/tmp/tts-venv/lib/python3.6/site-packages/TTS and then save the speech output in the same folder.

To achive that, what changes are required and which program i need to change to pass input sentence and then process the input text and save the output speech file. what is the python program i have to change for this .

[nayanhalder]

whatever i understand from server.py , i need to create a small program like this. am i correct.

from TTS.server.synthesizer import Synthesizer from scipy.io.wavfile import write text = " hi i am nayan" data = synthesizer.tts(text) sps = 44100 write("sample.wav", sps, data)

but i am getting an error File "/home/nayan/tmp/tts-venv/lib/python3.6/site-packages/scipy/io/wavfile.py", line 333, in write dkind = data.dtype.kind AttributeError: '_io.BytesIO' object has no attribute 'dtype'

[erogol]

closed due to inactivity

[Nakilon]

How do I install the pretrained models into Windows 10? Neither "windows" nor "install" are showing up on ctrl+F on your wiki page. I want the third-party application to use the voice other than the default one. It choses from the voices installed in OS.