[Released] Anilip 2 [Commercial]

1246789

Comments

  • DobitDobit Posts: 183

    I would like to warn you here to make these changes. The risk of something going wrong is relatively high. I also doubt that any of these new voices will work with Anilip 2 TextToSpeech as they are not SAPI 5 voices.

  • Thank you all for trying to help. I am not about to try hacking the system or messing with the registry. I have returned the product.

    For those of us not adept at figuring out all the technical jargon of that subscription service, perhaps a selection of voices to choose from in a future edition of Anilip would be possible? W10 will not let programs open their own folders in the main program folders, so too then these alternative voices [I suspect] would have to be stored elsewhere, like in a Dobit folder under Documents. Just an idea, along with a mike, and etc.

     

     

     

  • DobitDobit Posts: 183

    Thank you all for trying to help. I am not about to try hacking the system or messing with the registry. I have returned the product.

    For those of us not adept at figuring out all the technical jargon of that subscription service, perhaps a selection of voices to choose from in a future edition of Anilip would be possible? W10 will not let programs open their own folders in the main program folders, so too then these alternative voices [I suspect] would have to be stored elsewhere, like in a Dobit folder under Documents. Just an idea, along with a mike, and etc.

    I'm sorry you can't use Anilip 2. If I understand you correctly, that's because of the limited selection of Windows voices.

    We are aware of this problem for a while and have been testing third-party SAPI5 voices such as IVONA, Nuance or Cepstral. We will provide a free update of Anilip 2 in the next few weeks, in which we enable the integration of these voices. We will describe in the documentation which voices we can recommend with Anilip 2.

    We have a lot of plans for the new version 3 and it will take some time until it comes to the Daz shop. But one topic we're picking up here is a more user-friendly integration of cloud voices.

  • Dobit said:

    Thank you all for trying to help. I am not about to try hacking the system or messing with the registry. I have returned the product.

    For those of us not adept at figuring out all the technical jargon of that subscription service, perhaps a selection of voices to choose from in a future edition of Anilip would be possible? W10 will not let programs open their own folders in the main program folders, so too then these alternative voices [I suspect] would have to be stored elsewhere, like in a Dobit folder under Documents. Just an idea, along with a mike, and etc.

    I'm sorry you can't use Anilip 2. If I understand you correctly, that's because of the limited selection of Windows voices.

    We are aware of this problem for a while and have been testing third-party SAPI5 voices such as IVONA, Nuance or Cepstral. We will provide a free update of Anilip 2 in the next few weeks, in which we enable the integration of these voices. We will describe in the documentation which voices we can recommend with Anilip 2.

    We have a lot of plans for the new version 3 and it will take some time until it comes to the Daz shop. But one topic we're picking up here is a more user-friendly integration of cloud voices.

    More voices yes, after all in any group of people, each have their own voice. Cloud? well, I'd prefer on the computer. Maybe next edition. Thank you for trying.

     

  • Dobit said:

    To celebrate Anilip's 1 year anniversary, we will be offering a brand new upgrade in the next few days.
    So many new things have been added that we are upgrading to version 2.
    There was a long todo list we worked on last year.
    Here are just the most important innovations:

    1. Anilip 2 now supports many more voices.
      • "Speech Recognition" of voice recordings
      • "Text To Speech"
      • Cloud services (Transcribe and Polly from Amazon)
      • Generic JSON interface.
    2. A new viseme model allows a wide range of speech styles (from whispers to screams). Lip animators can now also implement emotional speech.
    3. Our new tabular lip animation tool makes it easy to translate even lengthly animations.
    4. The configuration of the lip animation was exposed for all figures in an XML file. Now also figures that are not offered by us can be adapted.
    5. We put a lot of work into documentation. 4 tutorial videos, 1 manual, inline help.

    A first impression is here.

     

    Mimic certainly could use a rest, and Anilip looks promising. I've installed it. At the moment I'm hung up on editing the dsx config file. I'm trying to add the new DS House Cat to the list of recognized files. I'm failing. If you could post a brief outline of how to go about it, I'll bet I would find what I'm overlooking right away. Thanks...

  • DobitDobit Posts: 183
    Dobit said:

    To celebrate Anilip's 1 year anniversary, we will be offering a brand new upgrade in the next few days.
    So many new things have been added that we are upgrading to version 2.
    There was a long todo list we worked on last year.
    Here are just the most important innovations:

    1. Anilip 2 now supports many more voices.
      • "Speech Recognition" of voice recordings
      • "Text To Speech"
      • Cloud services (Transcribe and Polly from Amazon)
      • Generic JSON interface.
    2. A new viseme model allows a wide range of speech styles (from whispers to screams). Lip animators can now also implement emotional speech.
    3. Our new tabular lip animation tool makes it easy to translate even lengthly animations.
    4. The configuration of the lip animation was exposed for all figures in an XML file. Now also figures that are not offered by us can be adapted.
    5. We put a lot of work into documentation. 4 tutorial videos, 1 manual, inline help.

    A first impression is here.

     

    Mimic certainly could use a rest, and Anilip looks promising. I've installed it. At the moment I'm hung up on editing the dsx config file. I'm trying to add the new DS House Cat to the list of recognized files. I'm failing. If you could post a brief outline of how to go about it, I'll bet I would find what I'm overlooking right away. Thanks...

    We will have a look to it and post an example.

  • DobitDobit Posts: 183
    Dobit said:
    Dobit said:

    To celebrate Anilip's 1 year anniversary, we will be offering a brand new upgrade in the next few days.
    So many new things have been added that we are upgrading to version 2.
    There was a long todo list we worked on last year.
    Here are just the most important innovations:

    1. Anilip 2 now supports many more voices.
      • "Speech Recognition" of voice recordings
      • "Text To Speech"
      • Cloud services (Transcribe and Polly from Amazon)
      • Generic JSON interface.
    2. A new viseme model allows a wide range of speech styles (from whispers to screams). Lip animators can now also implement emotional speech.
    3. Our new tabular lip animation tool makes it easy to translate even lengthly animations.
    4. The configuration of the lip animation was exposed for all figures in an XML file. Now also figures that are not offered by us can be adapted.
    5. We put a lot of work into documentation. 4 tutorial videos, 1 manual, inline help.

    A first impression is here.

     

    Mimic certainly could use a rest, and Anilip looks promising. I've installed it. At the moment I'm hung up on editing the dsx config file. I'm trying to add the new DS House Cat to the list of recognized files. I'm failing. If you could post a brief outline of how to go about it, I'll bet I would find what I'm overlooking right away. Thanks...

    We will have a look to it and post an example.

    We took the example of a house cat and wrote a small tutorial on how to integrate your own characters into Anilip 2. It was produced quickly, but it shows the essential steps.

    The tutorial can be viewed here.

    We uploaded an extended Anilip 2 configuration with an integration of the house cat. Just unzip and update the existing file.

    If you want to see other important DAZ figures integrated and don't want to do the configuration by yourself, just get in touch.

    zip
    zip
    dobitAnilip2.zip
    2K
  • If you want to use a voice that's not a computerized one, record it with any device you want and export it as a .wav file and you can get it to sync up. I recorded my actors' voices, exported the .wav through premiere, imported it into Anilip and it worked just fine. I tweaked the percentages and such, but it wasn't anything too difficult. Then I just exported the video and laid the video over the audio track and it sync'd up. 

    I'm sure if you don't have premiere, you can use davinci resolve to make a .wav, it's an editing program that is 100% free. 

    You can see what I did in this animated short I did. 

    https://youtu.be/H3bCkyLm47M

  • DobitDobit Posts: 183

    If you want to use a voice that's not a computerized one, record it with any device you want and export it as a .wav file and you can get it to sync up. I recorded my actors' voices, exported the .wav through premiere, imported it into Anilip and it worked just fine. I tweaked the percentages and such, but it wasn't anything too difficult. Then I just exported the video and laid the video over the audio track and it sync'd up. 

    I'm sure if you don't have premiere, you can use davinci resolve to make a .wav, it's an editing program that is 100% free. 

    You can see what I did in this animated short I did. 

    https://youtu.be/H3bCkyLm47M

    Thank you very much Bennie for this impressive example.

  • You're very welcome, Dobit, thanks for taking the time to watch and creating this invaluable asset for Daz! I'm getting ready to animate a feature film and I can't fathom the time I would spend hand keyframing the mouth movements for dialogue without Anilip 2.
  • Dobit said:

    We took the example of a house cat and wrote a small tutorial on how to integrate your own characters into Anilip 2. It was produced quickly, but it shows the essential steps.

    The tutorial can be viewed here.

    We uploaded an extended Anilip 2 configuration with an integration of the house cat. Just unzip and update the existing file.

    If you want to see other important DAZ figures integrated and don't want to do the configuration by yourself, just get in touch.

    This is good. I'm glad I came across this. I'm going to have a close look at it very soon. I would love to be able to do this. I would add that is is generous of you to offer this.

  • If you want to use a voice that's not a computerized one, record it with any device you want and export it as a .wav file and you can get it to sync up. I recorded my actors' voices, exported the .wav through premiere, imported it into Anilip and it worked just fine. I tweaked the percentages and such, but it wasn't anything too difficult. Then I just exported the video and laid the video over the audio track and it sync'd up. 

    I'm sure if you don't have premiere, you can use davinci resolve to make a .wav, it's an editing program that is 100% free. 

    You can see what I did in this animated short I did. 

    https://youtu.be/H3bCkyLm47M

    That's nice work.

  • James_HJames_H Posts: 1,031

    Very nice.

  • laststand@runbox.com[email protected] Posts: 866
    edited February 2020
    Dobit said:

    We took the example of a house cat and wrote a small tutorial on how to integrate your own characters into Anilip 2. It was produced quickly, but it shows the essential steps.

    The tutorial can be viewed here.

    We uploaded an extended Anilip 2 configuration with an integration of the house cat. Just unzip and update the existing file.

    If you want to see other important DAZ figures integrated and don't want to do the configuration by yourself, just get in touch.

    Wow. I plugged in your config file and it works great! You can't imagine how pleased I am. I thank you sincerely. This is the first time I've ever really seen Anilip at work. Now, I will learn how to use it. :-)

    Reallusion has animals. Hivewire has animals. Only DAZ though, to the best of my knowledge, has animals with visemes coded into them. Little use has been made of these. And now the DMC files are unavailable. About half a dozen DAZ animals have visemes. My wish is to learn enough to create a dbx file that includes all of these animals. Perhaps this may even be part of your upcoming release. ;-)

    I will now spend some time with the PDF Manual you provided. Good luck to me!

    .....................................................................

    Edit --

    OK. For me, finding the Node Name (YT Vid 1:36) was a breakthrough. This is probably where my own editing effort failed. I didn't know where to find this name so I guessed. Your demonstration of where to find the viseme name was also valuable. I have a bit confidence now in my ability to edit the config. Much obliged. Now to get familiar with the program itself. More later.

    Edit --

    Here are the results. I'm very happy with this. I tip my hat to Dobit.

    Post edited by [email protected] on
  • laststand@runbox.com[email protected] Posts: 866
    edited February 2020

    Moving along...

    I edited the config file to add the DAZ Milennium Cat. All went well to a point. Then I noticed that you sometimes needed to write "head:" to path to the viseme name. I did this. Hoo Ha! It worked like a charm!

    This is important to me Dobit. For one reason, the Mil Cat has quite a few pose and animation scripts available, but the House cat, so far, has none. For another, this will allow me to give voices to the other Milennium animals whose dmc config files are no longer available. I rarely have such good luck as this.

    Post edited by [email protected] on
  • laststand@runbox.com[email protected] Posts: 866
    edited February 2020
    Dobit said:
     

    We have a lot of plans for the new version 3...

    I'm an Anilip 2 fan now Dobit. Tell me something please.

    I happened to notice that the Anilip 2 product page lists 11 visemes titles 'Dog 8.'

    Does this mean that Dog 8 actually has its own visemes?? None are listed on the Dog 8 product page.

    :-)

    ...............................................................

    EDIT

    I now see that Dog 8 does indeed have visemes, and pops up immediately in Anilip 2. Thanks. I like this.

    Post edited by [email protected] on
  • laststand@runbox.com[email protected] Posts: 866
    edited February 2020

    -

    Post edited by [email protected] on
  • DobitDobit Posts: 183
    his x said:

    Moving along...

    I edited the config file to add the DAZ Milennium Cat. All went well to a point. Then I noticed that you sometimes needed to write "head:" to path to the viseme name. I did this. Hoo Ha! It worked like a charm!

    This is important to me Dobit. For one reason, the Mil Cat has quite a few pose and animation scripts available, but the House cat, so far, has none. For another, this will allow me to give voices to the other Milennium animals whose dmc config files are no longer available. I rarely have such good luck as this.

    The properties in the Anilip configuration refer to the root node if you do not specify anything else. If the properties are assigned to another node, the name of this node must be specified. In your case, the properties refer to the node named "head".

  • Dobit said:

    The properties in the Anilip configuration refer to the root node if you do not specify anything else. If the properties are assigned to another node, the name of this node must be specified. In your case, the properties refer to the node named "head".

    It works nicely. I have even realized that in your config file you identify all the current DAZ animals with visemes. At this time, the animal product pages don't always state that the animal has visemes. Dog  8 does not mention it. I have succeeded, after working through a few curious error messages, in using my own voice as source. The results were good even without a transcript. I am wondering if AWS JSON files can improve accuracy to any noticible degree.

     

  • laststand@runbox.com[email protected] Posts: 866
    edited February 2020

    -

    Post edited by [email protected] on
  • DobitDobit Posts: 183
    his x said:
    Dobit said:

    The properties in the Anilip configuration refer to the root node if you do not specify anything else. If the properties are assigned to another node, the name of this node must be specified. In your case, the properties refer to the node named "head".

    It works nicely. I have even realized that in your config file you identify all the current DAZ animals with visemes. At this time, the animal product pages don't always state that the animal has visemes. Dog  8 does not mention it. I have succeeded, after working through a few curious error messages, in using my own voice as source. The results were good even without a transcript. I am wondering if AWS JSON files can improve accuracy to any noticible degree.

     

    AWS Speech Recognition is somewhat more forgiving than that of Windows. But if the results are OK under Windows, I see no reason to work with AWS.

  • Dobit said:

    AWS Speech Recognition is somewhat more forgiving than that of Windows. But if the results are OK under Windows, I see no reason to work with AWS.

    I compared Anilip lypsync with AWS and with Mimic. I used the same recorded sound file each time. The sound file was recorded using two different mics, two different programs and two different computers. The results were not uniform. Since I am animating animals, I did not look for the level of accuracy that I would expect on a human character. Animal visemes are not always as convincing.

    • Mimic worked the first time each time
    • AWS JSON worked the first time each time
    • Anilip gave a lot of errors. Results were not consistent. See my attached screen grab.

     

    anilip error.jpg
    506 x 120 - 16K
  • DobitDobit Posts: 183
    his x said:
    Dobit said:

    AWS Speech Recognition is somewhat more forgiving than that of Windows. But if the results are OK under Windows, I see no reason to work with AWS.

    I compared Anilip lypsync with AWS and with Mimic. I used the same recorded sound file each time. The sound file was recorded using two different mics, two different programs and two different computers. The results were not uniform. Since I am animating animals, I did not look for the level of accuracy that I would expect on a human character. Animal visemes are not always as convincing.

    • Mimic worked the first time each time
    • AWS JSON worked the first time each time
    • Anilip gave a lot of errors. Results were not consistent. See my attached screen grab.

     

    Anilip uses Windows speech recognition by default. This requires a certain quality of the input signal. Microsoft recommends using a headset. Even better detection rates are achieved by entering the transcript. If you want to avoid the strict requirements, you can also use the AWS interface.

  • laststand@runbox.com[email protected] Posts: 866
    edited February 2020
    Dobit said:

    Anilip uses Windows speech recognition by default. This requires a certain quality of the input signal. Microsoft recommends using a headset. Even better detection rates are achieved by entering the transcript. If you want to avoid the strict requirements, you can also use the AWS interface.

    Thank you. Can you say where the certain quality is defined?

     

    Post edited by [email protected] on
  • DobitDobit Posts: 183
    his x said:
    Dobit said:

    Anilip uses Windows speech recognition by default. This requires a certain quality of the input signal. Microsoft recommends using a headset. Even better detection rates are achieved by entering the transcript. If you want to avoid the strict requirements, you can also use the AWS interface.

    Thank you. Can you say where the certain quality is defined?

     

    For best results we recommend:

    • 8 bits per sample
    • single channel mono
    • 22,050 samples per second
    • PCM encoding

    Your error message indicates that the level of your recording is too low. Speech recognition cannot distinguish the speech component from the noises. It often helps if you go closer to the microphone or use a headset. You can also "normalize" the level with audio software.

    Finally you can train the WIndows Speech Recognition. See Details here.

  • Dobit said:
    his x said:
    Dobit said:

    Anilip uses Windows speech recognition by default. This requires a certain quality of the input signal. Microsoft recommends using a headset. Even better detection rates are achieved by entering the transcript. If you want to avoid the strict requirements, you can also use the AWS interface.

    Thank you. Can you say where the certain quality is defined?

     

    For best results we recommend:

    • 8 bits per sample
    • single channel mono
    • 22,050 samples per second
    • PCM encoding

    Your error message indicates that the level of your recording is too low. Speech recognition cannot distinguish the speech component from the noises. It often helps if you go closer to the microphone or use a headset. You can also "normalize" the level with audio software.

    Finally you can train the WIndows Speech Recognition. See Details here.

    Excellent. Thank you. I will tend to this.

    I live in the desert. At night no cars go by for hours. The furnace is not running. I could hear a pin drop. I use a hand-held Blue Snowball. Audacity indicates the recording level is just short of distortion.

    My recordings have no problem in either Crazy Talk or in Mimic, so I assume this is primarily a Windows speech recognition issue. Do you agree?

     

  • DobitDobit Posts: 183
    his x said:
    Dobit said:
    his x said:
    Dobit said:

    Anilip uses Windows speech recognition by default. This requires a certain quality of the input signal. Microsoft recommends using a headset. Even better detection rates are achieved by entering the transcript. If you want to avoid the strict requirements, you can also use the AWS interface.

    Thank you. Can you say where the certain quality is defined?

     

    For best results we recommend:

    • 8 bits per sample
    • single channel mono
    • 22,050 samples per second
    • PCM encoding

    Your error message indicates that the level of your recording is too low. Speech recognition cannot distinguish the speech component from the noises. It often helps if you go closer to the microphone or use a headset. You can also "normalize" the level with audio software.

    Finally you can train the WIndows Speech Recognition. See Details here.

    Excellent. Thank you. I will tend to this.

    I live in the desert. At night no cars go by for hours. The furnace is not running. I could hear a pin drop. I use a hand-held Blue Snowball. Audacity indicates the recording level is just short of distortion.

    My recordings have no problem in either Crazy Talk or in Mimic, so I assume this is primarily a Windows speech recognition issue. Do you agree?

     

    You are welcome to send us examples of your recording, then we can look at it and say more.

  • Dobit said:

    You are welcome to send us examples of your recording, then we can look at it and say more.

    Thank you.

    I am assuming then that Anilip is unique in using the Windows STT system. Am I correct?

  • DobitDobit Posts: 183
    his x said:
    Dobit said:

    You are welcome to send us examples of your recording, then we can look at it and say more.

    Thank you.

    I am assuming then that Anilip is unique in using the Windows STT system. Am I correct?

    SAPI is a Windows standard that is used in countless products and solutions for speech recognition or Text2Speech since 1995. Alternatives to this can only be found in the cloud today. AWS, Google, Azure, IBM, Nuance and many more.

Sign In or Register to comment.