Install V‑Blaze 7.5

Note: This procedure must be performed as the root user and is intended for use with V‑Blaze version 7.5 or later on a RHEL/Oracle 8 or 9 system. For information about installing V‑Blaze version 7.4 or earlier, refer to .
Note: Installing on RHEL requires a RedHat subscription for access to the required packages. Microsoft Azure pay-as-you-go systems have the required RHEL packages installed automatically.
  1. Build a RHEL 8 or 9 system and ensure packages are updated to the latest versions. This system may be built on a cloud service such as AWS or Azure or on an on-premises managed server.
    Learn more: Installing RHEL 8 and Installing RHEL 9
  2. Update installed packages. 
    
dnf update

# Remove ol8 uek kernel as the default boot
dnf install kernel-headers kernel
grubby --set-default  $(grubby --info=ALL | grep ^kernel|grep -v uek|head -n1|cut -d'"' -f2)

# reboot
  3. Install the Voci repository on the new system. 
    
vociuser=<Voci-provided credentials>
vocipass=<Voci-provided credentials>
AUTH_TOKEN=<Voci-provided credentials>

echo "${vociuser}" >/etc/yum/vars/vociuser
echo "${vocipass}" >/etc/yum/vars/vocipass

dnf -y install https://${vociuser}:${vocipass}@repo.vocitec.com/main/asr/voci-repo-asr-7.5-2.noarch.rpm
dnf config-manager --add-repo https://repo.vocitec.com/public/voci-asr.repo
  4. Install EPEL twice. 
    
dnf -y install epel-release
dnf -y install epel-release
  5. Add the NVIDIA repository. 
    
# RHEL8
dnf config-manager --add-repo http://developer.download.nvidia.com/compute/cuda/repos/rhel8/x86_64/cuda-rhel8.repo

# RHEL 9
dnf config-manager --add-repo http://developer.download.nvidia.com/compute/cuda/repos/rhel9/x86_64/cuda-rhel9.repo

# Nvidia driver install
```
source /etc/os-release
distribution=rhel$(echo ${VERSION_ID%\.*})
ARCH=$( /bin/arch )
echo "dnf config-manager --add-repo http://developer.download.nvidia.com/compute/cuda/repos/$distribution/$ARCH/cuda-$distribution.repo" 2&gt;&amp;1 |tee -a 01.dkms.install.log
dnf config-manager --add-repo http://developer.download.nvidia.com/compute/cuda/repos/$distribution/$ARCH/cuda-$distribution.repo 2&gt;&amp;1 |tee -a 01.dkms.install.log
dnf -y install kernel-headers-$(uname -r) kernel-devel-$(uname -r) tar bzip2 make automake gcc gcc-c++ pciutils elfutils-libelf-devel libglvnd-opengl libglvnd-glx libglvnd-devel acpid pkgconfig dkms 2&gt;&amp;1 |tee -a 01.dkms.install.log
dnf -y module install nvidia-driver:latest-dkms 2&gt;&amp;1 |tee -a 01.dkms.install.log
# echo "blacklist nouveau" &gt; /etc/modprobe.d/blacklist-nouveau.conf
# echo &#39;omit_drivers+=" nouveau "&#39; &gt; /etc/dracut.conf.d/blacklist-nouveau.conf
# dracut --regenerate-all --force
depmod -a  2&gt;&amp;1 |tee -a 01.dkms.install.log
# secure boot will need the following after each kernel update.
# mokutil will prompt for you to make up a password, which will be used during reboot
# mokutil --import /var/lib/dkms/mok.pub  2&gt;&amp;1 |tee -a 01.dkms.install.log
 reboot
```
 # Check Nvidia driver is loaded
 ```
 # post reboot
 nvidia-smi  2&gt;&amp;1 |tee -a 01.dkms.install.log
  6. Apply your license before rebooting.
    Learn more: Enabling internet licensing
  7. Reboot the system to ensure that any kernel changes take effect.
    
dnf install -y voci-std-gpu11 voci-langpack-eng-us_callcenter voci-webapi.x86_64 

systemctl enable  vociserver

# Remember to apply your license before rebooting.
reboot
  8. Start up the core V‑Blaze services or restart them if they have already been started.
    This will take a bit of time as the language models you installed are loaded, and installation time varies by hardware specification. When the V‑Blaze ASR engine is ready to be used, log entries like the examples below appear in the log file /var/log/vociserver/err.log showing that the ASR engine is initializing and then ready. 
    2021-10-22 21:03:49.024619 initializing
2021-10-22 21:05:03.821191 serving on port 17170
  9. Configure language models.

    If you have more than one language model installed and want to manage how they are used, you can use the models.list file located in /opt/voci/. It contains options to configure language model sizing and availability parameters. A common use case for configuring models.list is to specify which models to load at startup and which models to load when requested (dynamic loading). Dynamic loading is available for language models only; all acoustic models load at startup.

    Learn more: Language model configuration

  10. Upgrade only the operating system once per month.
    Note: A restart is required after upgrading the system kernel.
    sudo dnf update --exclude voci-\*
  11. Update Voci software as needed.
    # By default the current release is best for production, and next generation repos hold the most current version used by Voci.
dnf update voci-\*

There is a known issue when installing Voci software with cURL on RHEL 8.9. Use the commands below to address the issue.


[FAILED] voci-repo-1.0.4-1.x86_64.rpm: Curl error (60): Peer certificate cannot be authenticated with given CA certificates for https://$username:$password/repo/private/user-acme/voci-repo-1.0.4-1.x86_64.rpm [SSL certificate problem: EE certificate key too weak]

# curl will not load the rpm from the voci repo
curl https://google.com

update-crypto-policies --show FUTURE

sudo update-crypto-policies --set DEFAULT

# issue corrected
curl https://google.com