diff --git a/Cargo.toml b/Cargo.toml
index 7402559..db31ee4 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -13,7 +13,6 @@ keywords = ["zkSNARKs", "cryptography", "proofs"]
 [dependencies]
 bellperson = { version = "0.24", default-features = false }
 ff = { version = "0.12.0", features = ["derive"] }
-merlin = {version = "3.0.0", default-features = false }
 digest = "0.8.1"
 sha3 = "0.8.2"
 rayon = "1.3.0"
@@ -40,6 +39,7 @@ pasta-msm = { version = "0.1.0", package = "lurk-pasta-msm" }
 [dev-dependencies]
 criterion = "0.3.1"
 rand = "0.8.4"
+hex = "0.4.3"
 
 [[bench]]
 name = "recursive-snark"
diff --git a/src/errors.rs b/src/errors.rs
index 2887d3d..8188e46 100644
--- a/src/errors.rs
+++ b/src/errors.rs
@@ -41,4 +41,7 @@ pub enum NovaError {
   /// returned when the step execution produces an output whose length differs from a previously declared arity
   #[error("InvalidStepOutputLength")]
   InvalidStepOutputLength,
+  /// returned when the transcript engine encounters an overflow of the round number
+  #[error("InternalTranscriptError")]
+  InternalTranscriptError,
 }
diff --git a/src/provider/ipa_pc.rs b/src/provider/ipa_pc.rs
index 8884475..d160a91 100644
--- a/src/provider/ipa_pc.rs
+++ b/src/provider/ipa_pc.rs
@@ -7,13 +7,12 @@ use crate::{
   traits::{
     commitment::{CommitmentEngineTrait, CommitmentGensTrait, CommitmentTrait},
     evaluation::EvaluationEngineTrait,
-    AppendToTranscriptTrait, ChallengeTrait, Group,
+    AppendToTranscriptTrait, ChallengeTrait, Group, TranscriptEngineTrait,
   },
   Commitment, CommitmentGens, CompressedCommitment, CE,
 };
 use core::{cmp::max, iter};
 use ff::Field;
-use merlin::Transcript;
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
 use std::marker::PhantomData;
@@ -58,7 +57,7 @@ where
 
   fn prove_batch(
     gens: &Self::EvaluationGens,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
     comms: &[Commitment<G>],
     polys: &[Vec<G::Scalar>],
     points: &[Vec<G::Scalar>],
@@ -89,7 +88,7 @@ where
         ),
         &InnerProductWitness::new(&polys[i]),
         transcript,
-      );
+      )?;
       nifs.push(n);
       r_U = u;
       r_W = w;
@@ -103,7 +102,7 @@ where
   /// A method to verify purported evaluations of a batch of polynomials
   fn verify_batch(
     gens: &Self::EvaluationGens,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
     comms: &[Commitment<G>],
     points: &[Vec<G::Scalar>],
     evals: &[G::Scalar],
@@ -129,7 +128,7 @@ where
           &evals[i],
         ),
         transcript,
-      );
+      )?;
       r_U = u;
       num_vars = max(num_vars, points[i].len());
     }
@@ -219,9 +218,9 @@ impl<G: Group> NIFSForInnerProduct<G> {
     W1: &InnerProductWitness<G>,
     U2: &InnerProductInstance<G>,
     W2: &InnerProductWitness<G>,
-    transcript: &mut Transcript,
-  ) -> (Self, InnerProductInstance<G>, InnerProductWitness<G>) {
-    transcript.append_message(b"protocol-name", Self::protocol_name());
+    transcript: &mut G::TE,
+  ) -> Result<(Self, InnerProductInstance<G>, InnerProductWitness<G>), NovaError> {
+    transcript.absorb_bytes(b"protocol-name", Self::protocol_name());
 
     // pad the instances and witness so they are of the same length
     let U1 = U1.pad(max(U1.b_vec.len(), U2.b_vec.len()));
@@ -230,21 +229,25 @@ impl<G: Group> NIFSForInnerProduct<G> {
     let W2 = W2.pad(max(U1.b_vec.len(), U2.b_vec.len()));
 
     // add the two commitments and two public vectors to the transcript
+    // we do not need to add public vectors as their compressed versions were
+    // read from the transcript
     U1.comm_a_vec
       .append_to_transcript(b"U1_comm_a_vec", transcript);
-    U1.b_vec.append_to_transcript(b"U1_b_vec", transcript);
     U2.comm_a_vec
       .append_to_transcript(b"U2_comm_a_vec", transcript);
-    U2.b_vec.append_to_transcript(b"U2_b_vec", transcript);
 
     // compute the cross-term
     let cross_term = inner_product(&W1.a_vec, &U2.b_vec) + inner_product(&W2.a_vec, &U1.b_vec);
 
     // add the cross-term to the transcript
-    cross_term.append_to_transcript(b"cross_term", transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &cross_term,
+      b"cross_term",
+      transcript,
+    );
 
     // obtain a random challenge
-    let r = G::Scalar::challenge(b"r", transcript);
+    let r = G::Scalar::challenge(b"r", transcript)?;
 
     // fold the vectors and their inner product
     let a_vec = W1
@@ -270,36 +273,38 @@ impl<G: Group> NIFSForInnerProduct<G> {
       c,
     };
 
-    (NIFSForInnerProduct { cross_term }, U, W)
+    Ok((NIFSForInnerProduct { cross_term }, U, W))
   }
 
   fn verify(
     &self,
     U1: &InnerProductInstance<G>,
     U2: &InnerProductInstance<G>,
-    transcript: &mut Transcript,
-  ) -> InnerProductInstance<G> {
-    transcript.append_message(b"protocol-name", Self::protocol_name());
+    transcript: &mut G::TE,
+  ) -> Result<InnerProductInstance<G>, NovaError> {
+    transcript.absorb_bytes(b"protocol-name", Self::protocol_name());
 
     // pad the instances so they are of the same length
     let U1 = U1.pad(max(U1.b_vec.len(), U2.b_vec.len()));
     let U2 = U2.pad(max(U1.b_vec.len(), U2.b_vec.len()));
 
     // add the two commitments and two public vectors to the transcript
+    // we do not need to add public vectors as their compressed representation
+    // were derived from the transcript
     U1.comm_a_vec
       .append_to_transcript(b"U1_comm_a_vec", transcript);
-    U1.b_vec.append_to_transcript(b"U1_b_vec", transcript);
     U2.comm_a_vec
       .append_to_transcript(b"U2_comm_a_vec", transcript);
-    U2.b_vec.append_to_transcript(b"U2_b_vec", transcript);
 
     // add the cross-term to the transcript
-    self
-      .cross_term
-      .append_to_transcript(b"cross_term", transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.cross_term,
+      b"cross_term",
+      transcript,
+    );
 
     // obtain a random challenge
-    let r = G::Scalar::challenge(b"r", transcript);
+    let r = G::Scalar::challenge(b"r", transcript)?;
 
     // fold the vectors and their inner product
     let b_vec = U1
@@ -311,11 +316,11 @@ impl<G: Group> NIFSForInnerProduct<G> {
     let c = U1.c + r * r * U2.c + r * self.cross_term;
     let comm_a_vec = U1.comm_a_vec + U2.comm_a_vec * r;
 
-    InnerProductInstance {
+    Ok(InnerProductInstance {
       comm_a_vec,
       b_vec,
       c,
-    }
+    })
   }
 }
 
@@ -343,27 +348,26 @@ where
     gens_c: &CommitmentGens<G>,
     U: &InnerProductInstance<G>,
     W: &InnerProductWitness<G>,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
   ) -> Result<Self, NovaError> {
-    transcript.append_message(b"protocol-name", Self::protocol_name());
+    transcript.absorb_bytes(b"protocol-name", Self::protocol_name());
 
     if U.b_vec.len() != W.a_vec.len() {
       return Err(NovaError::InvalidInputLength);
     }
 
     U.comm_a_vec.append_to_transcript(b"comm_a_vec", transcript);
-    U.b_vec.append_to_transcript(b"b_vec", transcript);
-    U.c.append_to_transcript(b"c", transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(&U.c, b"c", transcript);
 
     // sample a random base for commiting to the inner product
-    let r = G::Scalar::challenge(b"r", transcript);
+    let r = G::Scalar::challenge(b"r", transcript)?;
     let gens_c = gens_c.scale(&r);
 
     // a closure that executes a step of the recursive inner product argument
     let prove_inner = |a_vec: &[G::Scalar],
                        b_vec: &[G::Scalar],
                        gens: &CommitmentGens<G>,
-                       transcript: &mut Transcript|
+                       transcript: &mut G::TE|
      -> Result<
       (
         CompressedCommitment<G>,
@@ -402,7 +406,7 @@ where
       L.append_to_transcript(b"L", transcript);
       R.append_to_transcript(b"R", transcript);
 
-      let r = G::Scalar::challenge(b"challenge_r", transcript);
+      let r = G::Scalar::challenge(b"challenge_r", transcript)?;
       let r_inverse = r.invert().unwrap();
 
       // fold the left half and the right half
@@ -456,9 +460,9 @@ where
     gens_c: &CommitmentGens<G>,
     n: usize,
     U: &InnerProductInstance<G>,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
   ) -> Result<(), NovaError> {
-    transcript.append_message(b"protocol-name", Self::protocol_name());
+    transcript.absorb_bytes(b"protocol-name", Self::protocol_name());
     if U.b_vec.len() != n
       || n != (1 << self.L_vec.len())
       || self.L_vec.len() != self.R_vec.len()
@@ -468,11 +472,10 @@ where
     }
 
     U.comm_a_vec.append_to_transcript(b"comm_a_vec", transcript);
-    U.b_vec.append_to_transcript(b"b_vec", transcript);
-    U.c.append_to_transcript(b"c", transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(&U.c, b"c", transcript);
 
     // sample a random base for commiting to the inner product
-    let r = G::Scalar::challenge(b"r", transcript);
+    let r = G::Scalar::challenge(b"r", transcript)?;
     let gens_c = gens_c.scale(&r);
 
     let P = U.comm_a_vec + CE::<G>::commit(&gens_c, &[U.c]);
@@ -511,7 +514,7 @@ where
         self.R_vec[i].append_to_transcript(b"R", transcript);
         G::Scalar::challenge(b"challenge_r", transcript)
       })
-      .collect::<Vec<G::Scalar>>();
+      .collect::<Result<Vec<G::Scalar>, NovaError>>()?;
 
     // precompute scalars necessary for verification
     let r_square: Vec<G::Scalar> = (0..self.L_vec.len())
diff --git a/src/provider/keccak.rs b/src/provider/keccak.rs
new file mode 100644
index 0000000..c73cbca
--- /dev/null
+++ b/src/provider/keccak.rs
@@ -0,0 +1,160 @@
+//! This module provides an implementation of TranscriptEngineTrait using keccak256
+use crate::traits::PrimeFieldExt;
+use crate::{
+  errors::NovaError,
+  traits::{Group, TranscriptEngineTrait},
+};
+use core::marker::PhantomData;
+use sha3::{Digest, Keccak256};
+
+const PERSONA_TAG: &[u8] = b"NovaTranscript";
+const DOM_SEP_TAG: &[u8] = b"NovaRound";
+const KECCAK256_STATE_SIZE: usize = 64;
+const KECCAK256_PREFIX_CHALLENGE_LO: u8 = 0;
+const KECCAK256_PREFIX_CHALLENGE_HI: u8 = 1;
+
+/// Provides an implementation of TranscriptEngine
+#[derive(Debug, Clone)]
+pub struct Keccak256Transcript<G: Group> {
+  round: u16,
+  state: [u8; KECCAK256_STATE_SIZE],
+  transcript: Vec<u8>,
+  _p: PhantomData<G>,
+}
+
+fn compute_updated_state(input: &[u8]) -> [u8; KECCAK256_STATE_SIZE] {
+  let input_lo = [input, &[KECCAK256_PREFIX_CHALLENGE_LO]].concat();
+  let input_hi = [input, &[KECCAK256_PREFIX_CHALLENGE_HI]].concat();
+
+  let mut hasher_lo = Keccak256::new();
+  let mut hasher_hi = Keccak256::new();
+
+  hasher_lo.input(&input_lo);
+  hasher_hi.input(&input_hi);
+
+  let output_lo = hasher_lo.result();
+  let output_hi = hasher_hi.result();
+
+  [output_lo, output_hi]
+    .concat()
+    .as_slice()
+    .try_into()
+    .unwrap()
+}
+
+impl<G: Group> TranscriptEngineTrait<G> for Keccak256Transcript<G> {
+  fn new(label: &'static [u8]) -> Self {
+    let input = [PERSONA_TAG, label].concat();
+    let output = compute_updated_state(&input);
+
+    Self {
+      round: 0u16,
+      state: output,
+      transcript: vec![],
+      _p: Default::default(),
+    }
+  }
+
+  fn squeeze_scalar(&mut self, label: &'static [u8]) -> Result<G::Scalar, NovaError> {
+    let input = [
+      DOM_SEP_TAG,
+      self.round.to_le_bytes().as_ref(),
+      self.state.as_ref(),
+      self.transcript.as_ref(),
+      label,
+    ]
+    .concat();
+    let output = compute_updated_state(&input);
+
+    // update state
+    self.round = {
+      if let Some(v) = self.round.checked_add(1) {
+        v
+      } else {
+        return Err(NovaError::InternalTranscriptError);
+      }
+    };
+    self.state.copy_from_slice(&output);
+    self.transcript = vec![];
+
+    // squeeze out a challenge
+    Ok(G::Scalar::from_uniform(&output))
+  }
+
+  fn absorb_bytes(&mut self, label: &'static [u8], bytes: &[u8]) {
+    self.transcript.extend_from_slice(label);
+    self.transcript.extend_from_slice(bytes);
+  }
+}
+
+#[cfg(test)]
+mod tests {
+  use crate::{
+    provider::keccak::Keccak256Transcript,
+    traits::{AppendToTranscriptTrait, ChallengeTrait, Group, TranscriptEngineTrait},
+  };
+  use ff::PrimeField;
+  use sha3::{Digest, Keccak256};
+
+  type G = pasta_curves::pallas::Point;
+
+  #[test]
+  fn test_keccak_transcript() {
+    let mut transcript = Keccak256Transcript::new(b"test");
+
+    // two scalars
+    let s1 = <G as Group>::Scalar::from(2u64);
+    let s2 = <G as Group>::Scalar::from(5u64);
+
+    // add the scalars to the transcript
+    <<G as Group>::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &s1,
+      b"s1",
+      &mut transcript,
+    );
+    <<G as Group>::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &s2,
+      b"s2",
+      &mut transcript,
+    );
+
+    // make a challenge
+    let c1 =
+      <<G as Group>::Scalar as ChallengeTrait<G>>::challenge(b"challenge_c1", &mut transcript)
+        .unwrap();
+    assert_eq!(
+      hex::encode(c1.to_repr().as_ref()),
+      "51648083af5387a04a7aa2aec789ee78fdabe45dc1391d270a38fcb576447c01"
+    );
+
+    // a scalar
+    let s3 = <G as Group>::Scalar::from(128u64);
+
+    // add the scalar to the transcript
+    <<G as Group>::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &s3,
+      b"s3",
+      &mut transcript,
+    );
+
+    // make a challenge
+    let c2 =
+      <<G as Group>::Scalar as ChallengeTrait<G>>::challenge(b"challenge_c2", &mut transcript)
+        .unwrap();
+    assert_eq!(
+      hex::encode(c2.to_repr().as_ref()),
+      "9773f3349f7308153f6012e72b97fc304e48372bbd28bd122b37a8e46855d50f"
+    );
+  }
+
+  #[test]
+  fn test_keccak_example() {
+    let mut hasher = Keccak256::new();
+    hasher.input(0xffffffff_u32.to_le_bytes());
+    let output: [u8; 32] = hasher.result().try_into().unwrap();
+    assert_eq!(
+      hex::encode(output),
+      "29045a592007d0c246ef02c2223570da9522d0cf0f73282c79a1bc8f0bb2c238"
+    );
+  }
+}
diff --git a/src/provider/mod.rs b/src/provider/mod.rs
index c6d7a0d..ab55001 100644
--- a/src/provider/mod.rs
+++ b/src/provider/mod.rs
@@ -5,6 +5,7 @@
 //! `EvaluationEngine` with an IPA-based polynomial evaluation argument
 
 pub mod ipa_pc;
+pub mod keccak;
 pub mod pasta;
 pub mod pedersen;
 pub mod poseidon;
diff --git a/src/provider/pasta.rs b/src/provider/pasta.rs
index 4a47b9c..71ab3a4 100644
--- a/src/provider/pasta.rs
+++ b/src/provider/pasta.rs
@@ -1,23 +1,21 @@
 //! This module implements the Nova traits for pallas::Point, pallas::Scalar, vesta::Point, vesta::Scalar.
 use crate::{
   provider::{
+    keccak::Keccak256Transcript,
     pedersen::CommitmentEngine,
     poseidon::{PoseidonRO, PoseidonROCircuit},
   },
-  traits::{ChallengeTrait, CompressedGroup, Group},
+  traits::{CompressedGroup, Group, PrimeFieldExt},
 };
 use digest::{ExtendableOutput, Input};
-use ff::Field;
-use merlin::Transcript;
 use num_bigint::BigInt;
 use num_traits::Num;
 use pasta_curves::{
   self,
-  arithmetic::{CurveAffine, CurveExt, Group as OtherGroup},
+  arithmetic::{CurveAffine, CurveExt, FieldExt, Group as OtherGroup},
   group::{cofactor::CofactorCurveAffine, Curve, Group as AnotherGroup, GroupEncoding},
   pallas, vesta, Ep, EpAffine, Eq, EqAffine,
 };
-use rand_chacha::{rand_core::SeedableRng, ChaCha20Rng};
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
 use sha3::Shake256;
@@ -64,6 +62,7 @@ macro_rules! impl_traits {
       type PreprocessedGroupElement = $name::Affine;
       type RO = PoseidonRO<Self::Base, Self::Scalar>;
       type ROCircuit = PoseidonROCircuit<Self::Base>;
+      type TE = Keccak256Transcript<Self>;
       type CE = CommitmentEngine<Self>;
 
       #[cfg(any(target_arch = "x86_64", target_arch = "aarch64"))]
@@ -171,12 +170,10 @@ macro_rules! impl_traits {
       }
     }
 
-    impl ChallengeTrait for $name::Scalar {
-      fn challenge(label: &'static [u8], transcript: &mut Transcript) -> Self {
-        let mut key: <ChaCha20Rng as SeedableRng>::Seed = Default::default();
-        transcript.challenge_bytes(label, &mut key);
-        let mut rng = ChaCha20Rng::from_seed(key);
-        $name::Scalar::random(&mut rng)
+    impl PrimeFieldExt for $name::Scalar {
+      fn from_uniform(bytes: &[u8]) -> Self {
+        let bytes_arr: [u8; 64] = bytes.try_into().unwrap();
+        $name::Scalar::from_bytes_wide(&bytes_arr)
       }
     }
 
diff --git a/src/provider/pedersen.rs b/src/provider/pedersen.rs
index 14d79e2..2d0c5c8 100644
--- a/src/provider/pedersen.rs
+++ b/src/provider/pedersen.rs
@@ -6,6 +6,7 @@ use crate::{
       CommitmentEngineTrait, CommitmentGensTrait, CommitmentTrait, CompressedCommitmentTrait,
     },
     AbsorbInROTrait, AppendToTranscriptTrait, CompressedGroup, Group, ROTrait,
+    TranscriptEngineTrait,
   },
 };
 use core::{
@@ -13,8 +14,7 @@ use core::{
   marker::PhantomData,
   ops::{Add, AddAssign, Mul, MulAssign},
 };
-use ff::Field;
-use merlin::Transcript;
+use ff::{Field, PrimeField};
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
 
@@ -96,9 +96,17 @@ impl<C: CompressedGroup> CompressedCommitmentTrait<C> for CompressedCommitment<C
   }
 }
 
-impl<G: Group> AppendToTranscriptTrait for Commitment<G> {
-  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut Transcript) {
-    transcript.append_message(label, self.comm.compress().as_bytes());
+impl<G: Group> AppendToTranscriptTrait<G> for Commitment<G> {
+  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut G::TE) {
+    let (x, y, is_infinity) = self.comm.to_coordinates();
+    let is_infinity_byte = if is_infinity { 0u8 } else { 1u8 };
+    let bytes = [
+      x.to_repr().as_ref(),
+      y.to_repr().as_ref(),
+      &[is_infinity_byte],
+    ]
+    .concat();
+    transcript.absorb_bytes(label, &bytes);
   }
 }
 
@@ -115,9 +123,16 @@ impl<G: Group> AbsorbInROTrait<G> for Commitment<G> {
   }
 }
 
-impl<C: CompressedGroup> AppendToTranscriptTrait for CompressedCommitment<C> {
-  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut Transcript) {
-    transcript.append_message(label, self.comm.as_bytes());
+impl<C: CompressedGroup> AppendToTranscriptTrait<C::GroupElement> for CompressedCommitment<C> {
+  fn append_to_transcript(
+    &self,
+    label: &'static [u8],
+    transcript: &mut <C::GroupElement as Group>::TE,
+  ) {
+    let comm = self.decompress().unwrap();
+    <Commitment<C::GroupElement> as AppendToTranscriptTrait<C::GroupElement>>::append_to_transcript(
+      &comm, label, transcript,
+    );
   }
 }
 
diff --git a/src/r1cs.rs b/src/r1cs.rs
index a5a1b78..7dfad58 100644
--- a/src/r1cs.rs
+++ b/src/r1cs.rs
@@ -17,7 +17,6 @@ use core::cmp::max;
 use ff::Field;
 use flate2::{write::ZlibEncoder, Compression};
 use itertools::concat;
-use merlin::Transcript;
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
 use sha3::{Digest, Sha3_256};
@@ -440,11 +439,13 @@ impl<G: Group> R1CSShape<G> {
   }
 }
 
-impl<G: Group> AppendToTranscriptTrait for R1CSShape<G> {
-  fn append_to_transcript(&self, _label: &'static [u8], transcript: &mut Transcript) {
-    self
-      .get_digest()
-      .append_to_transcript(b"R1CSShape", transcript);
+impl<G: Group> AppendToTranscriptTrait<G> for R1CSShape<G> {
+  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut G::TE) {
+    <<G as Group>::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.get_digest(),
+      label,
+      transcript,
+    );
   }
 }
 
@@ -488,10 +489,10 @@ impl<G: Group> R1CSInstance<G> {
   }
 }
 
-impl<G: Group> AppendToTranscriptTrait for R1CSInstance<G> {
-  fn append_to_transcript(&self, _label: &'static [u8], transcript: &mut Transcript) {
+impl<G: Group> AppendToTranscriptTrait<G> for R1CSInstance<G> {
+  fn append_to_transcript(&self, _label: &'static [u8], transcript: &mut G::TE) {
     self.comm_W.append_to_transcript(b"comm_W", transcript);
-    self.X.append_to_transcript(b"X", transcript);
+    <[G::Scalar] as AppendToTranscriptTrait<G>>::append_to_transcript(&self.X, b"X", transcript);
   }
 }
 
@@ -629,12 +630,12 @@ impl<G: Group> RelaxedR1CSInstance<G> {
   }
 }
 
-impl<G: Group> AppendToTranscriptTrait for RelaxedR1CSInstance<G> {
-  fn append_to_transcript(&self, _label: &'static [u8], transcript: &mut Transcript) {
+impl<G: Group> AppendToTranscriptTrait<G> for RelaxedR1CSInstance<G> {
+  fn append_to_transcript(&self, _label: &'static [u8], transcript: &mut G::TE) {
     self.comm_W.append_to_transcript(b"comm_W", transcript);
     self.comm_E.append_to_transcript(b"comm_E", transcript);
-    self.u.append_to_transcript(b"u", transcript);
-    self.X.append_to_transcript(b"X", transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(&self.u, b"u", transcript);
+    <[G::Scalar] as AppendToTranscriptTrait<G>>::append_to_transcript(&self.X, b"X", transcript);
   }
 }
 
diff --git a/src/spartan/mod.rs b/src/spartan/mod.rs
index 195e251..b1c8dbb 100644
--- a/src/spartan/mod.rs
+++ b/src/spartan/mod.rs
@@ -9,12 +9,11 @@ use crate::{
   traits::{
     evaluation::EvaluationEngineTrait,
     snark::{ProverKeyTrait, RelaxedR1CSSNARKTrait, VerifierKeyTrait},
-    AppendToTranscriptTrait, ChallengeTrait, Group,
+    AppendToTranscriptTrait, ChallengeTrait, Group, TranscriptEngineTrait,
   },
 };
 use ff::Field;
 use itertools::concat;
-use merlin::Transcript;
 use polynomial::{EqPolynomial, MultilinearPolynomial, SparsePolynomial};
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
@@ -82,7 +81,7 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
     U: &RelaxedR1CSInstance<G>,
     W: &RelaxedR1CSWitness<G>,
   ) -> Result<Self, NovaError> {
-    let mut transcript = Transcript::new(b"RelaxedR1CSSNARK");
+    let mut transcript = G::TE::new(b"RelaxedR1CSSNARK");
 
     // sanity check that R1CSShape has certain size characteristics
     assert_eq!(pk.S.num_cons.next_power_of_two(), pk.S.num_cons);
@@ -105,7 +104,7 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
     // outer sum-check
     let tau = (0..num_rounds_x)
       .map(|_i| G::Scalar::challenge(b"challenge_tau", &mut transcript))
-      .collect();
+      .collect::<Result<Vec<G::Scalar>, NovaError>>()?;
 
     let mut poly_tau = MultilinearPolynomial::new(EqPolynomial::new(tau).evals());
     let (mut poly_Az, mut poly_Bz, poly_Cz, mut poly_uCz_E) = {
@@ -136,22 +135,39 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
       &mut poly_uCz_E,
       comb_func_outer,
       &mut transcript,
-    );
+    )?;
 
     // claims from the end of sum-check
     let (claim_Az, claim_Bz): (G::Scalar, G::Scalar) = (claims_outer[1], claims_outer[2]);
-
-    claim_Az.append_to_transcript(b"claim_Az", &mut transcript);
-    claim_Bz.append_to_transcript(b"claim_Bz", &mut transcript);
     let claim_Cz = poly_Cz.evaluate(&r_x);
+
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &claim_Az,
+      b"claim_Az",
+      &mut transcript,
+    );
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &claim_Bz,
+      b"claim_Bz",
+      &mut transcript,
+    );
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &claim_Cz,
+      b"claim_Cz",
+      &mut transcript,
+    );
+
     let eval_E = MultilinearPolynomial::new(W.E.clone()).evaluate(&r_x);
-    claim_Cz.append_to_transcript(b"claim_Cz", &mut transcript);
-    eval_E.append_to_transcript(b"eval_E", &mut transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &eval_E,
+      b"eval_E",
+      &mut transcript,
+    );
 
     // inner sum-check
-    let r_A = G::Scalar::challenge(b"challenge_rA", &mut transcript);
-    let r_B = G::Scalar::challenge(b"challenge_rB", &mut transcript);
-    let r_C = G::Scalar::challenge(b"challenge_rC", &mut transcript);
+    let r_A = G::Scalar::challenge(b"challenge_rA", &mut transcript)?;
+    let r_B = G::Scalar::challenge(b"challenge_rB", &mut transcript)?;
+    let r_C = G::Scalar::challenge(b"challenge_rC", &mut transcript)?;
     let claim_inner_joint = r_A * claim_Az + r_B * claim_Bz + r_C * claim_Cz;
 
     let poly_ABC = {
@@ -219,10 +235,14 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
       &mut MultilinearPolynomial::new(poly_z),
       comb_func,
       &mut transcript,
-    );
+    )?;
 
     let eval_W = MultilinearPolynomial::new(W.W.clone()).evaluate(&r_y[1..]);
-    eval_W.append_to_transcript(b"eval_W", &mut transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &eval_W,
+      b"eval_W",
+      &mut transcript,
+    );
 
     let eval_arg = EE::prove_batch(
       &pk.gens,
@@ -245,7 +265,7 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
 
   /// verifies a proof of satisfiability of a RelaxedR1CS instance
   fn verify(&self, vk: &Self::VerifierKey, U: &RelaxedR1CSInstance<G>) -> Result<(), NovaError> {
-    let mut transcript = Transcript::new(b"RelaxedR1CSSNARK");
+    let mut transcript = G::TE::new(b"RelaxedR1CSSNARK");
 
     // append the R1CSShape and RelaxedR1CSInstance to the transcript
     vk.S.append_to_transcript(b"S", &mut transcript);
@@ -259,7 +279,7 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
     // outer sum-check
     let tau = (0..num_rounds_x)
       .map(|_i| G::Scalar::challenge(b"challenge_tau", &mut transcript))
-      .collect::<Vec<G::Scalar>>();
+      .collect::<Result<Vec<G::Scalar>, NovaError>>()?;
 
     let (claim_outer_final, r_x) =
       self
@@ -275,24 +295,31 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
       return Err(NovaError::InvalidSumcheckProof);
     }
 
-    self
-      .claims_outer
-      .0
-      .append_to_transcript(b"claim_Az", &mut transcript);
-    self
-      .claims_outer
-      .1
-      .append_to_transcript(b"claim_Bz", &mut transcript);
-    self
-      .claims_outer
-      .2
-      .append_to_transcript(b"claim_Cz", &mut transcript);
-    self.eval_E.append_to_transcript(b"eval_E", &mut transcript);
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.claims_outer.0,
+      b"claim_Az",
+      &mut transcript,
+    );
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.claims_outer.1,
+      b"claim_Bz",
+      &mut transcript,
+    );
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.claims_outer.2,
+      b"claim_Cz",
+      &mut transcript,
+    );
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.eval_E,
+      b"eval_E",
+      &mut transcript,
+    );
 
     // inner sum-check
-    let r_A = G::Scalar::challenge(b"challenge_rA", &mut transcript);
-    let r_B = G::Scalar::challenge(b"challenge_rB", &mut transcript);
-    let r_C = G::Scalar::challenge(b"challenge_rC", &mut transcript);
+    let r_A = G::Scalar::challenge(b"challenge_rA", &mut transcript)?;
+    let r_B = G::Scalar::challenge(b"challenge_rB", &mut transcript)?;
+    let r_C = G::Scalar::challenge(b"challenge_rC", &mut transcript)?;
     let claim_inner_joint =
       r_A * self.claims_outer.0 + r_B * self.claims_outer.1 + r_C * self.claims_outer.2;
 
@@ -346,7 +373,11 @@ impl<G: Group, EE: EvaluationEngineTrait<G, CE = G::CE>> RelaxedR1CSSNARKTrait<G
     }
 
     // verify eval_W and eval_E
-    self.eval_W.append_to_transcript(b"eval_W", &mut transcript); //eval_E is already in the transcript
+    <G::Scalar as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.eval_W,
+      b"eval_W",
+      &mut transcript,
+    ); //eval_E is already in the transcript
 
     EE::verify_batch(
       &vk.gens,
diff --git a/src/spartan/sumcheck.rs b/src/spartan/sumcheck.rs
index deb29f0..d32d07b 100644
--- a/src/spartan/sumcheck.rs
+++ b/src/spartan/sumcheck.rs
@@ -5,7 +5,6 @@ use crate::errors::NovaError;
 use crate::traits::{AppendToTranscriptTrait, ChallengeTrait, Group};
 use core::marker::PhantomData;
 use ff::Field;
-use merlin::Transcript;
 use rayon::prelude::*;
 use serde::{Deserialize, Serialize};
 
@@ -21,7 +20,7 @@ impl<G: Group> SumcheckProof<G> {
     claim: G::Scalar,
     num_rounds: usize,
     degree_bound: usize,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
   ) -> Result<(G::Scalar, Vec<G::Scalar>), NovaError> {
     let mut e = claim;
     let mut r: Vec<G::Scalar> = Vec::new();
@@ -48,7 +47,7 @@ impl<G: Group> SumcheckProof<G> {
       poly.append_to_transcript(b"poly", transcript);
 
       //derive the verifier's challenge for the next round
-      let r_i = G::Scalar::challenge(b"challenge_nextround", transcript);
+      let r_i = G::Scalar::challenge(b"challenge_nextround", transcript)?;
 
       r.push(r_i);
 
@@ -65,8 +64,8 @@ impl<G: Group> SumcheckProof<G> {
     poly_A: &mut MultilinearPolynomial<G::Scalar>,
     poly_B: &mut MultilinearPolynomial<G::Scalar>,
     comb_func: F,
-    transcript: &mut Transcript,
-  ) -> (Self, Vec<G::Scalar>, Vec<G::Scalar>)
+    transcript: &mut G::TE,
+  ) -> Result<(Self, Vec<G::Scalar>, Vec<G::Scalar>), NovaError>
   where
     F: Fn(&G::Scalar, &G::Scalar) -> G::Scalar + Sync,
   {
@@ -103,7 +102,7 @@ impl<G: Group> SumcheckProof<G> {
       poly.append_to_transcript(b"poly", transcript);
 
       //derive the verifier's challenge for the next round
-      let r_i = G::Scalar::challenge(b"challenge_nextround", transcript);
+      let r_i = G::Scalar::challenge(b"challenge_nextround", transcript)?;
       r.push(r_i);
       polys.push(poly.compress());
 
@@ -115,13 +114,13 @@ impl<G: Group> SumcheckProof<G> {
       poly_B.bound_poly_var_top(&r_i);
     }
 
-    (
+    Ok((
       SumcheckProof {
         compressed_polys: polys,
       },
       r,
       vec![poly_A[0], poly_B[0]],
-    )
+    ))
   }
 
   pub fn prove_cubic_with_additive_term<F>(
@@ -132,8 +131,8 @@ impl<G: Group> SumcheckProof<G> {
     poly_C: &mut MultilinearPolynomial<G::Scalar>,
     poly_D: &mut MultilinearPolynomial<G::Scalar>,
     comb_func: F,
-    transcript: &mut Transcript,
-  ) -> (Self, Vec<G::Scalar>, Vec<G::Scalar>)
+    transcript: &mut G::TE,
+  ) -> Result<(Self, Vec<G::Scalar>, Vec<G::Scalar>), NovaError>
   where
     F: Fn(&G::Scalar, &G::Scalar, &G::Scalar, &G::Scalar) -> G::Scalar + Sync,
   {
@@ -195,7 +194,7 @@ impl<G: Group> SumcheckProof<G> {
       poly.append_to_transcript(b"poly", transcript);
 
       //derive the verifier's challenge for the next round
-      let r_i = G::Scalar::challenge(b"challenge_nextround", transcript);
+      let r_i = G::Scalar::challenge(b"challenge_nextround", transcript)?;
       r.push(r_i);
       polys.push(poly.compress());
 
@@ -209,13 +208,13 @@ impl<G: Group> SumcheckProof<G> {
       poly_D.bound_poly_var_top(&r_i);
     }
 
-    (
+    Ok((
       SumcheckProof {
         compressed_polys: polys,
       },
       r,
       vec![poly_A[0], poly_B[0], poly_C[0], poly_D[0]],
-    )
+    ))
   }
 }
 
@@ -322,12 +321,12 @@ impl<G: Group> CompressedUniPoly<G> {
   }
 }
 
-impl<G: Group> AppendToTranscriptTrait for UniPoly<G> {
-  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut Transcript) {
-    transcript.append_message(label, b"UniPoly_begin");
-    for i in 0..self.coeffs.len() {
-      self.coeffs[i].append_to_transcript(b"coeff", transcript);
-    }
-    transcript.append_message(label, b"UniPoly_end");
+impl<G: Group> AppendToTranscriptTrait<G> for UniPoly<G> {
+  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut G::TE) {
+    <[G::Scalar] as AppendToTranscriptTrait<G>>::append_to_transcript(
+      &self.coeffs,
+      label,
+      transcript,
+    );
   }
 }
diff --git a/src/traits/commitment.rs b/src/traits/commitment.rs
index 0a10f4f..e4c4dc9 100644
--- a/src/traits/commitment.rs
+++ b/src/traits/commitment.rs
@@ -71,7 +71,7 @@ pub trait CommitmentTrait<G: Group>:
   + Serialize
   + for<'de> Deserialize<'de>
   + AbsorbInROTrait<G>
-  + AppendToTranscriptTrait
+  + AppendToTranscriptTrait<G>
   + CommitmentOps
   + CommitmentOpsOwned
   + ScalarMul<G::Scalar>
@@ -96,7 +96,7 @@ pub trait CompressedCommitmentTrait<C: CompressedGroup>:
   + Sync
   + Serialize
   + for<'de> Deserialize<'de>
-  + AppendToTranscriptTrait
+  + AppendToTranscriptTrait<C::GroupElement>
 {
   /// Holds the type of the commitment that can be decompressed into
   type Commitment;
diff --git a/src/traits/evaluation.rs b/src/traits/evaluation.rs
index 431ec96..6853faf 100644
--- a/src/traits/evaluation.rs
+++ b/src/traits/evaluation.rs
@@ -5,7 +5,6 @@ use crate::{
   errors::NovaError,
   traits::{commitment::CommitmentEngineTrait, Group},
 };
-use merlin::Transcript;
 use serde::{Deserialize, Serialize};
 
 /// A trait that ties different pieces of the commitment evaluation together
@@ -27,7 +26,7 @@ pub trait EvaluationEngineTrait<G: Group>:
   /// A method to prove evaluations of a batch of polynomials
   fn prove_batch(
     gens: &Self::EvaluationGens,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
     comm: &[<Self::CE as CommitmentEngineTrait<G>>::Commitment],
     polys: &[Vec<G::Scalar>],
     points: &[Vec<G::Scalar>],
@@ -37,7 +36,7 @@ pub trait EvaluationEngineTrait<G: Group>:
   /// A method to verify purported evaluations of a batch of polynomials
   fn verify_batch(
     gens: &Self::EvaluationGens,
-    transcript: &mut Transcript,
+    transcript: &mut G::TE,
     comm: &[<Self::CE as CommitmentEngineTrait<G>>::Commitment],
     points: &[Vec<G::Scalar>],
     evals: &[G::Scalar],
diff --git a/src/traits/mod.rs b/src/traits/mod.rs
index 98251b8..9ba9136 100644
--- a/src/traits/mod.rs
+++ b/src/traits/mod.rs
@@ -1,4 +1,5 @@
 //! This module defines various traits required by the users of the library to implement.
+use crate::errors::NovaError;
 use bellperson::{
   gadgets::{boolean::AllocatedBit, num::AllocatedNum},
   ConstraintSystem, SynthesisError,
@@ -8,7 +9,6 @@ use core::{
   ops::{Add, AddAssign, Mul, MulAssign, Sub, SubAssign},
 };
 use ff::{PrimeField, PrimeFieldBits};
-use merlin::Transcript;
 use num_bigint::BigInt;
 use serde::{Deserialize, Serialize};
 
@@ -39,7 +39,8 @@ pub trait Group:
   /// A type representing an element of the scalar field of the group
   type Scalar: PrimeField
     + PrimeFieldBits
-    + ChallengeTrait
+    + PrimeFieldExt
+    + ChallengeTrait<Self>
     + Send
     + Sync
     + Serialize
@@ -53,13 +54,16 @@ pub trait Group:
   /// A type representing preprocessed group element
   type PreprocessedGroupElement: Clone + Debug + Send + Sync + Serialize + for<'de> Deserialize<'de>;
 
-  /// A type that represents a hash function that consumes elements
+  /// A type that represents a circuit-friendly sponge that consumes elements
   /// from the base field and squeezes out elements of the scalar field
   type RO: ROTrait<Self::Base, Self::Scalar> + Serialize + for<'de> Deserialize<'de>;
 
   /// An alternate implementation of Self::RO in the circuit model
   type ROCircuit: ROCircuitTrait<Self::Base> + Serialize + for<'de> Deserialize<'de>;
 
+  /// A type that provides a generic Fiat-Shamir transcript to be used when externalizing proofs
+  type TE: TranscriptEngineTrait<Self>;
+
   /// A type that defines a commitment engine over scalars in the group
   type CE: CommitmentEngineTrait<Self> + Serialize + for<'de> Deserialize<'de>;
 
@@ -105,24 +109,12 @@ pub trait CompressedGroup:
   fn as_bytes(&self) -> &[u8];
 }
 
-/// A helper trait to append different types to the transcript
-pub trait AppendToTranscriptTrait {
-  /// appends the value to the transcript under the provided label
-  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut Transcript);
-}
-
 /// A helper trait to absorb different objects in RO
 pub trait AbsorbInROTrait<G: Group> {
   /// Absorbs the value in the provided RO
   fn absorb_in_ro(&self, ro: &mut G::RO);
 }
 
-/// A helper trait to generate challenges using a transcript object
-pub trait ChallengeTrait {
-  /// Returns a Scalar representing the challenge using the transcript
-  fn challenge(label: &'static [u8], transcript: &mut Transcript) -> Self;
-}
-
 /// A helper trait that defines the behavior of a hash function that we use as an RO
 pub trait ROTrait<Base, Scalar> {
   /// A type representing constants/parameters associated with the hash function
@@ -204,17 +196,63 @@ impl<T, Rhs, Output> ScalarMul<Rhs, Output> for T where T: Mul<Rhs, Output = Out
 pub trait ScalarMulOwned<Rhs, Output = Self>: for<'r> ScalarMul<&'r Rhs, Output> {}
 impl<T, Rhs, Output> ScalarMulOwned<Rhs, Output> for T where T: for<'r> ScalarMul<&'r Rhs, Output> {}
 
-impl<F: PrimeField> AppendToTranscriptTrait for F {
-  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut Transcript) {
-    transcript.append_message(label, self.to_repr().as_ref());
+/// This trait defines the behavior of a transcript engine compatible with Spartan
+pub trait TranscriptEngineTrait<G: Group>: Send + Sync {
+  /// initializes the transcript
+  fn new(label: &'static [u8]) -> Self;
+
+  /// returns a scalar element of the group as a challenge
+  fn squeeze_scalar(&mut self, label: &'static [u8]) -> Result<G::Scalar, NovaError>;
+
+  /// absorbs a label and a sequence of bytes
+  fn absorb_bytes(&mut self, label: &'static [u8], bytes: &[u8]);
+}
+
+/// A helper trait to append different types to the transcript
+pub trait AppendToTranscriptTrait<G: Group> {
+  /// appends the value to the transcript under the provided label
+  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut G::TE);
+}
+
+/// A helper trait to generate challenges using a transcript object
+pub trait ChallengeTrait<G: Group> {
+  /// Returns a challenge from the transcript
+  fn challenge(label: &'static [u8], transcript: &mut G::TE) -> Result<Self, NovaError>
+  where
+    Self: Sized;
+}
+
+/// Defines additional methods on PrimeField objects
+pub trait PrimeFieldExt: PrimeField {
+  /// Returns a Scalar representing the bytes
+  fn from_uniform(bytes: &[u8]) -> Self;
+
+  /// Returns a byte representation
+  fn to_bytes(v: &[Self]) -> Vec<u8> {
+    (0..v.len())
+      .map(|i| v[i].to_repr().as_ref().to_vec())
+      .collect::<Vec<Vec<u8>>>()
+      .into_iter()
+      .flatten()
+      .collect::<Vec<u8>>()
+  }
+}
+
+impl<G: Group<Scalar = F>, F: PrimeField> ChallengeTrait<G> for F {
+  fn challenge(label: &'static [u8], transcript: &mut G::TE) -> Result<F, NovaError> {
+    transcript.squeeze_scalar(label)
+  }
+}
+
+impl<G: Group<Scalar = F>, F: PrimeField> AppendToTranscriptTrait<G> for F {
+  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut G::TE) {
+    transcript.absorb_bytes(label, self.to_repr().as_ref());
   }
 }
 
-impl<F: PrimeField> AppendToTranscriptTrait for [F] {
-  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut Transcript) {
-    for s in self {
-      s.append_to_transcript(label, transcript);
-    }
+impl<G: Group<Scalar = F>, F: PrimeField + PrimeFieldExt> AppendToTranscriptTrait<G> for [F] {
+  fn append_to_transcript(&self, label: &'static [u8], transcript: &mut G::TE) {
+    transcript.absorb_bytes(label, &<F as PrimeFieldExt>::to_bytes(self));
   }
 }